The Wind and Beyond of History at Auburn University, Has Written About Aerospace History for the Past 25 Years

About the editors: James R. Hansen, Professor The Wind and Beyond of History at Auburn University, has written about aerospace history for the past 25 years. His newest A Documentary Journey into the History book, First Man: The Life of Neil A. Armstrong (Simon & TheWind and The and of Aerodynamics in America Schuster, 2005), provides the authorized and definitive Beyond Wind biography of the famous test pilot, astronaut, and first Volume II: Reinventing the Airplane man on the Moon. His two-volume study of NASA The airplane ranks as one of history’s most inge- Langley Research Center—Engineer in Charge (NASA SP- Beyond nious and phenomenal inventions. It has surely been 4305, 1987) and Spaceflight Revolution (NASA SP-4308, A Documentary Journey A Documentary Journey into the one of the most world changing. How ideas about 1995) earned significant critical acclaim. His other into the History of aerodynamics first came together and how the science books include From the Ground Up (Smithsonian, 1988), History of Aerodynamics in America and technology evolved to forge the airplane into the Enchanted Rendezvous (NASA Monographs in Aerospace Aerodynamics revolutionary machine that it became is the epic story History #4, 1995), and The Bird Is On The Wing (Texas in America told in this six-volume series, The Wind and Beyond: A A&M University Press, 2003). Documentary Journey through the History of Aerodynamics in Jeremy R. Kinney is a curator in the Aeronautics America. Division, National Air and Space Museum, Smithsonian Institution and holds a Ph.D. in the history of technology Following up on Volume I’s account of the invention from Auburn. of the airplane and the creation of the original aero- D. Bryan Taylor is a doctoral candidate in the history nautical research establishment in the United States, of technology at Auburn. Volume II Volume II explores the airplane design revolution of Molly Prickett holds a B.A. degree in English from the 1920s and 1930s and the quest for improved air- Auburn University. foils. Subsequent volumes cover the aerodynamics of J. Lawrence Lee holds B.M.E., M.A., and Ph.D. airships, flying boats, rotary-wing aircraft, breaking the degrees from Auburn. He is an engineer-historian at sound barrier, and more. the Historic American Engineering Record, a branch of the U.S. National Park Service, and is former chair of In 2005, the Society for the History of Technology the History and Heritage Committee of the American awarded its first annual Eugene S. Ferguson Prize for Society of Mechanical Engineering. outstanding and original reference works to The Wind and Beyond. The citation read in part:

“The Wind and Beyond is remarkable in its breadth of Cover: Fluid Dynamics, Tina York. The study of vision. Its purview includes not just aerodynamical the- fluid dynamics attempts to explain what happens ories and research results, but also innovative airships to an object when it encounters the friction of at- and airship components as well as the institutions in mospheric resistance (such as a plane encounter- which and through which aerodynamics developed… ing resistance as it speeds through the air). The art- James R. Hansen, Each [chapter] essay is original in two ways. First, each is ist has decided to depict the effect of air flow as a Editor a first-rate piece of scholarship in its own right. Second, plane or other flying objects move through the air. the very decision to include these narratives is signifi- NASA Image 95-HC-379. with Jeremy Kinney, cant: they comprise roughly 10 percent of the contents D. Bryan Taylor, of the volume, but they make the other 90 percent both accessible and meaningful to the nonspecialist reader, Molly Prickett, and simultaneously enhancing the value of and enlarging J. Lawrence Lee the potential audience for the whole volume….The Wind and Beyond will be a boon both to students and to established scholars in several ways. Like many similar collections, it provides one-stop access to documents The NASA History Series that were previously scattered in many different places. Volume II: Reinventing the Airplane Going beyond other similar collections, however, The National Aeronautics and Space Administration Wind and Beyond makes the documents intellectually as NASA History Division well as physically accessible…The end result is an emi- Office of External Relations James R. Hansen, Editor nently readable reference work, one that is truly, as its title suggests, the beginning of a journey rather than Washington, D.C. with Jeremy Kinney, D. Bryan Taylor, Molly Prickett, and J. Lawrence Lee the end.” 2007 NASA SP-2007-4409 The Wind and Beyond

Vol. II NASA SP-2007-4409

The Wind and Beyond: A Documentary Journey into the History of Aerodynamics in America

Volume II: Reinventing the Airplane

James R. Hansen, Editor with Jeremy Kinney, D. Bryan Taylor, Molly F. Prickett, and J. Lawrence Lee

The NASA History Series

National Aeronautics and Space Administration NASA History Division Office of External Relations Washington, DC

Table of Contents

Foreword xxiii Acknowledgments xxv Introduction to Volume II xxvii Reinventing the Airplane xxvii Biographies of Volume II Contributors xxxiii

Chapter Three: The Design Revolution 1

Essay: The Design Revolution in Aircraft Aerodynamics 1 The Documents 47 3-1, Excerpts from Frederick W. Lanchester, Aerodynamics: Constituting the First Volume of a Complete Work on Aerial Flight (New York: D. Van Nostrand Company, 1908) 47 3-2, Ludwig Prandtl, “Motion of Fluids with Very Little Viscosity,” NACA Technical Memorandum No. 452 Washington, DC, 1928. 71 3-3(a), Alexander Klemin, Chapter X: “Resistance of Various Airplane Parts”. 80 3-3(b), Grover C. Loening, Military Aeroplanes: An Explanatory Consid- eration of their Characteristics, Performances, Construction, Mainte- nance, and Operation (Boston: W.S. Best, 1918) . 87 3-4, Louis Breguet, “Aerodynamical Efficiency and the Reduction of Air Transport Costs,” Aeronautical Journal (August 1922) . 102 Library of Congress Cataloging-in-Publication Data 3-5, Captain W. H. Sayers, “The Lesson of the Schneider Cup Race,” Aviation 15 (5 November 1923) 119 The Wind and Beyond: A Documentary Journey into the History of 3-6, John K. Northrop, quoted in Garry R. Pape, Northrop Flying Wings: Aerodynamics in America/ James R. Hansen, editor; with D. Bryan Taylor, Jeremy A History of Jack Northrop’s Visionary Aircraft (Atglen, Penn.: Schiffer, Kinney, Molly Prickett, and J. Lawrence Lee. 1995) 128 3-7, B. Melville Jones, “The Streamline Aeroplane,” Aeronautical Journal p. cm. -- (The NASA History Series) (NASA SP-2007-4409) 33 (1929) 129 3-8, Excerpts from Fred E. Weick and James R. Hansen, From the Includes bibliographical references and index. Ground Up: The Autobiography of an Aeronautical Engineer (Wash- Contents: v. 2. Reinventing the Airplane ington and London: Smithsonian Institution Press, 1988) 164 1. Aerodynamics--Research--United States--History. 2. Aeronautical 3-9(a), Western Union telegram, Jerry Vultee, Lockheed Aircraft Co., engineers--United States. I. Hansen, James R. II. Series III. Series: Burbank, CA, to NACA, “Attention Lieutenant Tom Carroll,” 5 NASA SP-4409 Feb. 1929, copy in NACA Research Authorization file 215, NASA Langley Historical Archives . 176 TL570.W64 2007 3-9(b), Fred E. Weick, “The New NACA Low-Drag Cowling,” Aviation 25 (17 November 1928) 177 2002035774 3-10(a), H. C. H. Townend, “Reduction of Drag of Radial Engine by 3-15(b), Hal L. Hibbard, “Problems in Fast Air-Transport Design,” Attachment of Rings of Aerofoil Section,” British Aeronautical Mechanical Engineering 55 (October 1933) 285 Research Committee Research and Memoranda 1267 (1929). 184 3-15(c), “Preliminary Study of Retractable Landing Gears for High and 3-10(b), United Aircraft and Transport Corporation Technical Advisory Low Wing Monoplanes,” Air Corps Information Circular 7 (18 Feb- Committee, Meeting Minutes, December 1929, Boeing Historical ruary 1933) 289 Archives, Seattle, Washington 203 3-15(d), Richard M. Mock, “Retractable Landing Gears,”Aviation 32 3-10(c), “The Curtiss Anti-Drag Ring,” Curtiss-Wright Review 1 (Decem- (February 1933) 302 ber 1930) 225 3-15(e), Roy G. Miller, letter to editor, Aviation, “Retractable Landing 3-11(a), United Aircraft and Transport Corporation Technical Advisory Gears,” Aviation 32 (April 1933), with response from Richard M. Committee Minutes held at the Pratt & Whitney plant, Hartford, Mock 311 CT, 19-23 May 1930. 228 3-16(a), Fred E. Weick of Hampton Virginia, U.S. Patent 2,110,516. 3-11(b), “The Reminiscences of Harold Hicks,” August 1951, transcript, Filed 18 Jan. 1934; issued 8 March 1938 321 pp. 61-69. Henry Ford Museum and Greenfield Village Research 3-16(b), Fred E. Weick & Associates, Inc., to Waco Aircraft Co., Troy Center, Dearborn, Mich. 232 OH, 8 Apr. 1938 340 3-12(a), Harlan D. Fowler to Rueben H. Fleet, 8 January 1929,Folder 3-16(c), Weick & Associates, Inc., to Waco Aircraft Co., Troy, OH, 2 “Fowler Flap,” copy in San Diego Aerospace Museum, CA. 238 Feb. 1939. 340 3-12(b), Consolidated Aircraft Corporation to Harlan D. Fowler, 21 3-16(d), C. J. Brukner, President, The Waco Aircraft Company, Troy, January 1929, Folder “Fowler Flap,” copy in San Diego Aerospace OH, to Fred E. Weick and Associates, Inc., Evelyn Place, College Museum, CA. 241 Park, MD, 16 Feb. 1939 341 3-12(c), Harlan D. Fowler, “Variable Lift,” Western Flying 10 (November 3-16(e), Weick & Associates, Inc., to The Waco Aircraft Company, Troy, 1931) 242 OH, Attn: Mr. C. J. Brukner, President, 22 March 1939. 342 3-12(d), Harlan D. Fowler to William B. Mayo, 15 August 1932, copy 3-16(f), Weick & Associates, to Timm Aircraft Corporation, Grand in Accession 18, Box 96, Folder “Fowler Variable Wing,” Henry Central Air Terminal, 1020 Airway, Glendale, CA, Attn: Mr. R. A. Ford Museum and Greenfield Village Research Center . 250 Powell, Vice-President, 22 March 1939. 343 3-12(e), H.A. Hicks to Harlan D. Fowler, 10 September 1932,copy in 3-16(g), Alpheus Barnes, The Wright Company, 11 Pine Street,New Accession 18, Box 96, Folder “Fowler Variable Wing,” Henry Ford York, to Mr. Grover C. Loening, c/o U.S.S. Mississippi, Pensacola, Museum and Greenfield Village Research Center . 251 FL, 16 Jan. 1915 344 3-12(f), T.P. Wright, “The Application of Slots and Flaps to Airplane 3-16(h), Grover Loening, “What Might Have Been,” in Our Wings Grow Wings in America,” U.S. Air Services (September 1933) 252 Faster (1935). 344 3-12(g), Robert C. Platt, “Aerodynamic Characteristics of a Wing with 3-17(a), A. L. Klein, “Effect of Fillets on Wing-Fuselage Interference,” Fowler Flaps including Flap Loads, Downwash, and Calculated Transactions of the American Society of Mechanical Engineers 56 Effecton Take-off,” NACA Technical Report 534 (Washington, (1934) 355 1935). 257 3-17(b), James A. White and Manly J. Hood, “Wing-Fuselage Interfer- 3-12(h), Harlan D. Fowler, “Aerodynamic Characteristics of the Fowler ence, Tail Buffeting, and Air Flow About the Tail of a Low-Wing- Wing,” Aero Digest 29 (September 1936) 267 Monoplane,” NACA Technical Report 482 (Washington, 1934). 367 3-13, Standard Steel Propeller Company to Ryan Airlines, undated (ca. 3-18(a), Shatswell Ober, “Report on Wind Tunnel Model Tests, Ford Tri- April 1927), in Charles A. Lindbergh, The Spirit of St. Louis (New Motor Plane—4-AT,” 17 February 1927 396 York: Scribner, 1953; reprint, St. Paul: Minnesota Historical Society Press, 1993) . 276 3-18(b), Aircraft Engineering Department, Ford Motor Company, 3-14, C. B. Allen, “Hamilton Standard Wins Collier Trophy for Control- “Recent Design Changes on the 5-AT and their Effect on High lable Pitch Propeller,” The Bee-Hive 8 (June 1934) 277 Speed,” undated (ca. 1929) 403 3-15(a), Smith J. DeFrance, “The Aerodynamic Effect of a Retractable 3-18(c), “The Reminiscences of Harold Hicks,” Aug. 1951. 407 Landing Gear,” NACA Technical Note 456 (Washington, 1933) 283 3-19, “Lockheed ‘Sirius’ Monoplane,” Aero Digest 16 (January 1930). 412 3-20(a), “The New Martin Bomber: Model XB-907,” Glenn L. Martin 3-27(b), H. J. E. Reid, Engineer-in-Charge, to NACA, “Program pro- Company Engineering Report No. 240, 5 August 1932 417 posed by Mr. Rhode for determination of dynamic overstress of 3-20(b), Glenn L. Martin to Walter G. Kilner, undated (ca. 1933) 427 wings in gusts,” March 15, 1933 524 3-20(c), Glenn L. Martin Company, “History and Development of the 3-27(c), G.W. Lewis, Director of Aeronautical Research, to LMAL, Martin Bomber,” 20 January 1938 428 “Research authorization on study of load and load type distribution 3-21, Jack Frye to Douglas Aircraft Corporation, “Attention: Mr. on commercial type airplane,” April 5, 1933 . 525 Donald Douglas,” 2 August 1932. A facsimile of this letter appeared 3-27(d), Richard V. Rhode, Aeronautical Engineer, to Engineer-in- in American Heritage of Invention and Technology, Fall 1988 434 Charge, “Proposed daily weather flights and gust tail load measure- 3-22(a), Clark B. Millikan and Arthur L. Klein,“Report on Wind Tunnel ments on modern airplane,” June 6, 1935. 526 Test on a Model of the Douglas Transport DC-1 with Various 3-27(e), Extracts from letter dated April 12, 1936 from Edward P. Modifications” 438 Warner to Dr. J. S. Ames, National Advisory Committee for Aero- 3-22(b), Engineering Department, Douglas Aircraft Company, “Devel- nautics, with comments of Dr. D.M. Little, United States Weather opment of the Douglas Transport,” Technical Data Report SW- Bureau. 528 157A, undated (ca. 1933-34). 455 3-27(f), Edward P. Warner to Dr. J. S. Ames, National Advisory Com- 3-23, “Douglas Airliner for Transcontinental Service,” Aviation 32 (Octo- mittee for Aeronautics, April 18, 1936 . 531 ber 1933). 461 3-27(g), C. W. Lewis, Director of Aeronautical Research, to Mr. L.V. 3-24, Donald W. Douglas, Sr., “The Development and Reliability of the- Kerber, Chief, Aircraft Airworthiness Section, Bureau of Air Com- Modern Multi-Engine Air Liner,” The Journal of the Royal Aeronau- merce, February 26, 1938. 532 tical Society (November 1935) 465 3-27(h), Interview of Richard V. Rhode by Michael D. Keller, January 10, 3-25(a), George W. Lewis, Director of Aeronautical Research, NACA, to 1967. 534 Mr. R. D. Kelly, Research Supervisor, United Air Lines Transport 3-28(a), Floyd L. Thompson, Associate Aeronautical Engineer, to Engi- Corporation, Chicago, IL, 3 Sept. 1937. 482 neer-in-Charge [Henry J. E. Reid], NACA Langley, “Preliminary 3-25(b), John W. Crowley, Jr., Senior Aeronautical Engineer, to Engi- study of control requirements for large transport airplanes,” 14 July neer-in-Charge [Henry J. E. Reid], Langley Memorial Aeronautical 1936. 540 Laboratory, Langley Field, VA, 5 Oct. 1937. 483 3-28(b), Edward P. Warner to Mr. Leighton W. Rogers, Aeronautical 3-25(c), R. D. Kelly, Research Supervisor, United Air Lines Transport Chamber of Commerce, Shoreham Building, Washington, DC, 19 Corporation, Field Headquarters, 6 Oct. 1937. 485 Nov. 1936, copy in Research Authorization (RA) file 509, LHA 547 3-25(d), Melvin N. Gough, Senior Test Pilot, Langley Memorial Aero- 3-28(c), William H. Hernnstein, Jr., Assistant Aeronautical Engineer, to nautical Laboratory, Langley Field, VA, to Engineer-in-Charge, Engineer-in-Charge, NACA Langley, “Visit of Mr. E. P. Warner to “Suggestion that conference on ‘Stalling’ be held with interested the Laboratory,” December 3, 1936 549 personnel outside the Laboratory,” 2 March 1938 487 3-28(d), Eastman N. Jacobs, Aeronautical Engineer, to Engineer-in- 3-25(e), Edward P. Warner, New York City, to Dr. George W. Lewis, Charge, NACA Langley, “Mr. E. P. Warner’s visit to the Laboratory NACA, 6 Apr. 1938. 488 to discuss airplane tests—December 3, 1936” 551 3-26(a), Excerpts from Lewis A. Rodert and Alun R. Jones,“Profile Drag 3-28(e), Hartley A. Soulé, Assistant Aeronautical Engineer, to Engineer- Investigations of an Airplane Wing Equipped with Rubber Inflat- in-Charge, NACA Langley, “Visit of Mr. Warner to flight section able De-Icer,” NACA Advance Confidential Report, Dec. 1939. 495 on June 14, 1937,” 15 June 1937. 552 3-26(b), Excerpts from George W. Gray, Chapter 14,“Heat Against Ice,” 3-28(f), Floyd L. Thompson, Associate Aeronautical Engineer, to Engi- in Frontiers of Flight: The Story of NACA Research (New York: Alfred neer-in-Charge, NACA Langley, “Tests of flying qualities of Martin A. Knopf, 1948). 506 M-10B airplane,” 14 July 1937. 553 3-27(a), Richard V. Rhode, Associate Aeronautical Engineer, to Mr. 3-28(g), Edward P. Warner, 11 West 42nd Street, New York City, to Dr. Charles Parker, Secretary, Aeronautical Chamber of Commerce, George W. Lewis, NACA, Navy Building, Washington, DC, 28 “Load Factors,” April 14, 1932 522 Dec. 1937. 556 3-28(h), Robert R. Gilruth, “Requirements for Satisfactory Flying Chapter four: on the wing 657 Qualities of Airplanes,” NACA Advanced Confidential Report, Apr. 1941. 558 Essay: The Design Revolution in Aircraft Aerodynamics 3-29(a), C. H. Dearborn, “Full-Scale Wind-Tunnel Tests of Navy XF2A- The Documents 695 1 Fighter Airplane,” NACA Confidential Memorandum Report, 17 4-1, Edgar S. Gorrell and H.S. Martin, Aerofoils andAerofoil Structural May 1938. 578 Combinations, NACA Technical Report 18 (Washington, 1918) 695 3-29(b), H. J. E. Reid, Engineer-in-Charge, LMAL, to NACA, “Drag- 4-2, Colonel Virginius E. Clark, “Design Your Own Airfoils,” Aviation reduction investigation on XP-40 airplane,” 7 Apr. 1939. 582 (Oct. 1927) 703 3-29(c), Don R. Berlin, Chief Engineer, Curtiss Aeroplane Division, 4-3, Max M. Munk, “General Theory of Thin Wing Sections,” NACA Curtiss Wright Corp., to Major Carl F. Greene, Materiel Division, Technical Report 142 (Washington, 1922) 712 Liaison Officer, NACA Laboratory, 4 Aug. 1939 584 4-4, Max M. Munk and Elton W. Miller, “Model Tests with a Systematic 3-29(d), C. H. Dearborn, Aeronautical Engineer, LMAL, to Engineer-in- Series of 27 Wing Sections at Full Reynolds Number,” NACA Tech- Charge, “Investigation for determining means of increasing the high nical Report 221 (Washington, 1925) 715 speed of the XP-41 airplane,” 8 June 1939. 586 4-5(a), R.H. Upson to the NACA, “Attention: Mr. Victory,” 19 Novem- 3-30(a), General Henry H. Arnold to NACA, “Full-Scale Wind-Tunnel ber 1928, in Research Authorization (RA) file 270, Historical Tests of XP-39,” 9 June 1939 591 Archives, NASA Langley Research Center, Hampton, Va. 723 3-30(b), Smith J. DeFrance to Chief, Aerodynamics Division, Langley 4-5(b), Eastman N. Jacobs to Elton W. Miller, “Suggestions from Mr. R. Aeronautical Laboratory, “Estimated High-Speed of the XP-39 H. Upson,” undated (ca. 1 February 1929), RA file 270 724 Airplane,” 25 Aug. 1939. 592 4-5(c), R. H. Upson to Dr. G.W. Lewis, 19 March 1929, RA file 270 725 3-30(c), Larry Bell, President, Bell Aircraft, to George W. Lewis, NACA, 4-5(d), Eastman N. Jacobs to Elton W. Miller, 4 April 1929. 726 17 Jan. 1940. 593 4-5(e), G. W. Lewis to Langley Memorial Aeronautical Laboratory, “Air- 3-31(a), Excerpts from Clinton H. Dearborn, “The Effect of Rivet Heads foil tests suggested by Mr. R.H. Upson,” 22 April 1929 727 on the Characteristics of a 6 By 36 Foot Clark Y Metal Airfoil,” 4-5(f), Eastman N. Jacobs to Chief of the Aerodynamics Division (Elton NACA Technical Note 461, May 1933 596 W. Miller), “Airfoil testing program in Variable Density Tunnel,” 17 3-31(b), Excerpts from Manley J. Hood, “The Effects of Some Common May 1929, RA file 270. 728 Surface Irregularities on Wing Drag,” NACA Technical Note 695, 4-6, Eastman N. Jacobs, Kenneth E. Ward, and Robert M. Pinkerton, March 1939. 605 “The Characteristics of 78 Related Airfoil Sections from Tests in the 3-31(c), Charles Peyton Autry, Boeing Aircraft Company, “Drag of Riv- Variable-density Wind Tunnel,” NACA Technical Report 460 (Wash- eted Wings,” in Aviation (May 1941) 618 ington, 1933). 730 3-32, J. D. Akerman for the Boeing Aircraft Company, excerpts from 4-7(a), C.W. Howard, Major, Air Corps, and Chief, Engineering Section, Manual for Design of Details Affecting Aerodynamical Characteristics Materiel Division, Office of the Chief of the Division, Wright Field, of an Airplane, Boeing Report No. D-2789, 12 Sept. 1940 624 Dayton, Oh., 13 Jan. 1933, in RA file 290, LHA, Hampton, Va. 739 3-33(a), Laurence K. Loftin, Jr. Chap. 6, “Design Maturity, 1945-1980,” 4-7(b), Montgomery Knight, Director, Daniel Guggenheim School of in Quest for Performance: The Evolution of Modern Aircraft Aeronautics, Georgia School of Technology, Atlanta, Georgia, to (Washington, DC: NASA SP-468, 1985). 640 George W. Lewis, NACA, Navy Building, Washington, D.C., 19 3-33(b), Laurence K. Loftin, Jr. Chap. 6, “Design Trends,” in Quest for January 1932, in Research Authorization (RA) file 290, LHA, Performance: The Evolution of Modern Aircraft (Washington, DC: Hampton, Va. 740 NASA SP-468, 1985) 648 4-7(c), Edward P. Warner, Editor, Aviation, New York, N.Y., to George W. Lewis, NACA, 20 January 1933, in RA file 290, LHA, Hamp- ton, Va. 741 4-7(d), G.W. Lewis to Dr. Joseph Ames, Johns Hopkins University, Balti- more, Md., 10 June 1933, in RA file 290, LHA. 742 4-7(e), “Possible Saving by Use of N.A.C.A. 2415 Airfoil,” undated (ca. 4-20, Eastman N. Jacobs, Senior Aeronautical Engineer, “Investigation of summer 1932), in RA file 290, LHA 743 low-drag airfoil sections having extensive laminar boundary layers,” 4-8(a), Charles H. Chatfield, Research Division, United Aircraft Corpo- undated (but typed 27 June 1938) 840 ration, East Hartford, Connecticut, to Dr. G.W. Lewis, NACA, 8 4-21, Eastman N. Jacobs, Senior Aeronautical Engineer, to Chief of the April 1937, in Research Authorization (RA) file 290, LHA, Hamp- Aerodynamics Division, “Patent on airfoil developments,” 9 Decem- ton, Va. 746 ber 1938 845 4-8(b), G.W. Lewis to Charles H. Chatfield, United Aircraft Corpora- 4-22, Newton H. Anderson, Aircraft Layout and Detail Design, 1941 849 tion, 1 May 1937, RA file 290, LHA 747 4-23(a), Eastman N. Jacobs to Engineer-In-Charge, “Conversation with 4-9(a), G.W. Lewis, NACA,To The Members of the Board of Award, Dr. Lewis regarding application of laminar-flow airfoils,” February Sylvanus Albert Reed Award for 1937 751 1939. 857 4-9(b), R.C. Platt, Memorandum for Dr. Lewis, “Airfoil sections 4-23(b), H.J.E. Reid, Engineer-in-Charge, LMAL, to NACA, “Aerody- employed for wings of modern airplanes,” 2 September 1937, RA namically smooth finishes for airplanes—information for Vultee file 290, LHA, Hampton Va. 754 Aircraft, Inc.,” November 1940. 858 4-10, H.J.E. Reid, Engineer-in-Charge, Langley Field, to NACA, “Paper 4-23(c), G.W. Lewis, Director of Aeronautical Research, to the Chairman entitled ‘A Few Present Problems in Aerodynamics,’ by Dr. von of the NACA (Dr. Joseph Ames), “Investigation of laminar-flow Kármán,” 8 February 1933 756 low-drag wings,” November 1939 861 4-11, Eastman N. Jacobs, Memorandum to Engineer-in-Charge, “Scale 4-23(d), G.W. Lewis, Director of Aeronautical Research, to Langley effect on airfoils, conference,” 11 April 1934,” in RA file 88,LHA, Memorial Aeronautical Laboratory, “Investigation in flight of lami- Hampton, Va. 760 nar-flow low-drag wings,” January 1940 862 4-12, G.W. Lewis, Director of Aeronautical Research, NACA, to Mr. 4-24(a), Robert J. Woods, Chief Design Engineer, Bell Aircraft Corp., J.L. Naylor, Secretary, Aeronautical Research Committee, National 2050 Elmwood Avenue, Buffalo NY, to NACA, “Attn: Dr. George Physical Laboratory, Teddington, Middlesex, England, 13 April W. Lewis,” March 1940 865 1937, in RA file 290, LHA, Hampton, Va. 764 4-24(b), G.W. Lewis, Director of Aeronautical Research to Langley 4-13, Hugh B. Freeman to Chief of the Aerodynamics Division [Elton Memorial Aeronautical Laboratory, “Request for information on W. Miller], “Boundary-layer research,” 18 April 1932, in RA file new airfoil sections—Bell Aircraft Corporation,” March 1940. 866 201,LHA, Hampton, Va. 769 4-24(c), H.J.E. Reid, Engineer-in-Charge, to NACA, “Request for infor- 4-14, Theodore Theodorsen, Senior Physicist, to Engineer-in-Charge, mation on new airfoil sections—Bell Aircraft Corporation,” March “Boundary layer removal,” 4 Feb. 1932, in RA 88, LHA, Hampton, 1940. 867 Va. 773 4-24(d), G.W. Lewis, Director of Aeronautical Research, to Robert J. 4-15, Theodore Theodorsen, Introduction,“Theory of Wing Sections of Woods, Chief Design Engineer, Bell Aircraft Corporation, March Arbitrary Shape,” NACA Technical Report 411 (1931) 776 1940. 868 4-16, Eastman Jacobs, Memorandum to Engineer-in-Charge,“Program 4-24(e), Eastman N. Jacobs to Engineer-in-Charge, “Visit of Mr. Robert for study of scale effect on airfoils,” 13 November 1929, in RA file J. Woods of Bell Aircraft to LMAL, February 1941”. 869 88, LHA, Hampton, Va. 783 4-25(a), Eastman N. Jacobs, Principal Aeronautical Engineer, Memo- 4-17, B. Melvill Jones, “Flight Experiments on the Boundary Layer,” randum to Director of Aeronautical Research (George W. Lewis), Journal of the Aeronautical Sciences 5, January 1938 786 “Application of new airfoil data to experimental military airplanes,” 4-18, Eastman N. Jacobs, Senior Aeronautical Engineer,to Engineer-In- June 1940. 873 Charge, “Notes on the history of the development of the laminar- 4-25(b), George W. Lewis, Director of Aeronautical Research, NACA, to flow airfoils and on the range of shapes included,” 27 December Chief of the Bureau of Aeronautics, Navy Department, Washington, 1938, copy in LHA, Hampton, VA. 823 D.C., “Confidential memorandum report regarding the application 4-19, Eastman N. Jacobs, “Laminar and Turbulent Boundary Layers as of new airfoil-section data of the laminar-flow type for current and Affecting Practical Aerodynamics,” Journal of the Society of Automo- new airplane designs,” 17 June 1940. 874 tive Engineers 40, March 1937 828 4-25(c), W. H. Herrnstein, Aeronautical Engineer (LMAL), to Engineer- in-Charge, “Discussion between Dr. G. W. Lewis and members of L. Lindemann, Air Attache, British Embassy, Washington, D.C., Laboratory staff relative to airfoil selection problems,” July 1940 . 875 December 1940. 909 4-25(d), Edwin P. Hartman, Western Coordinating Officer of the NACA, 4-28(b), G.W. Lewis, Director of Aeronautical Research, NACA, to Sir Santa Monica, CA., to Coordinator of Research, “Visit to Ryan- Henry Tizard, Chairman, Aeronautical Research Committee, c/o Aeronautical Company,” August 1940 877 Director of Intelligence, Air Ministry, London, W.C. 2, England, 4-25(e), Eastman N. Jacobs, Principal Aeronautical Engineer, to Engi- January 1941. 910 neer-in-Charge, “Visit to the Buffalo Curtis plant at the request of 4-28(c), H.T. Tizard, Ministry of Aircraft Production, Millbank, to Dr. Don Berlin, September 1940” 878 George Lewis, Director of Research, NACA, February 1941. 911 4-25(f), Arthur E. Raymond, Vice President of Engineering, Douglas 4-28(d), G. W. Lewis to Tizard (via British Embassy for Diplomatic Aircraft Company, Inc., Santa Monica, CA, to Dr. G.W. Lewis, Pouch), March 1941. 912 Director of Aeronautical Research, NACA, March 1941 880 4-28(e) Edward Warner, American Embassy, 1 Grosvenor Square, 4-25(g), Elton W. Miller, Chief of the Aerodynamics Division (LMAL) London, to Dr. George W. Lewis, NACA 1500 New Hampshire to Engineer-in-Charge, “Visit of Mr. L. C. Miller of the Brewster Ave., Washington, D.C., August 1942. 914 Company to the Laboratory on January 3, 1941,” 4 January 1941, 4-28(f) Eastman N. Jacobs, Principal Aeronautical Engineer, NACA RA file 290 881 Langley, to Engineer-in-Charge, “Low-drag airfoils in England,” 4-26(a), George J. Mead, Director, Airplane and Engine Division, September 1942. 916 National Defense Council, Federal Reserve Building, to Aeronauti- 4-28(g), Ivan H. Driggs, Aviation Design Research Branch, Bureau of cal Board, Room 1907, Navy Building, August 1940. 885 Aeronautics, Navy Department, Washington, D.C., to Experiments 4-26(b), Ira C. Eaker, Lieutenant Colonel, Air Corps, For the Chief and Developments Branch, BuAer, “British Views upon Airfoils for of the Air Corps, to Dr. George J. Mead, Director, Airplane and High Speed Aircraft,” August 1943. 917 Engine Division, National Defense Council, Federal Reserve Build- 4-28(h), Ira H. Abbott, Senior Aeronautical Engineer, to Engineer-in- ing, September 1940. 886 Charge, “British views upon airfoils for high-speed aircraft,” 4-26(c), G.W. Lewis, Director of Aeronautical Research, NACA, to September 1943. 921 Edward J. Horkey, Aerodynamics Department, North American 4-29, Edwin P. Hartman, NACA West Coast Coordinating Office,to Aviation, Inc., Inglewood, CA, 1 November 1940. 887 Chief of Research Coordination, NACA, “Information from the 4-26(d), Excerpts from North American Aviation, Inc., Manufacturing industry on the application of low-drag airfoils,” CONFIDEN- Division, Engineering Department, Inglewood CA, “Aerodynamic TIAL, July 1944. 925 Load Calculations for Model NA-73 Airplane,” North American 4-30(a), Theodore von Kármán, “Laminar Flow Wings,” in Where We Report NA-5041, March 1941. 890 Stand (written in 1945 and issued as an Army Air Forces Report in 4-27(a), Edward Warner, Civil Aeronautics Board, Washington, D.C., to 1946). 936 Eastman N. Jacobs, Principal Aeronautical Engineer, NACA Lang- 4-30(b), von Kármán, “Aerodynamic Problems,” in Science, The Key to ley, October 1940. 900 Air Supremacy (originally published in 1945). 939 4-27(b), Eastman N. Jacobs, Principal Aeronautical Engineer, NACA Langley, to Dr. Edward Warner, NACA, Navy Building, Washing- NASA History Series 943 ton, D.C., October 1940. 901 4-27(c), Eastman N. Jacobs, Principal Aeronautical Engineer,to Dr. Index 953 Edward P. Warner, Chairman, Committee on Aerodynamics, NACA, Navy Building, Washington, D.C., November 1940. 902 4-27(d), G.W. Lewis to LMAL, “Report by Mr. Jacobs on present status of laminar-flow wing development, for next meeting of the Aerody- namics Committee, to be held the latter part of January,” December 1940 906 4-28(a), Vannevar Bush, Chairman, NACA, to Brigadier General C.

Foreword

Airplane travel is surely one of the most significant technological achievements of the last century. The impact of the airplane goes far beyond the realm of the history of technology and touches upon virtually every aspect of society from eco- nomics to politics to engineering and science. While space exploration often claims more public glory than aeronautics research, many more individuals have been able to fly within the Earth’s atmosphere than above it. Thus aeronautics and air travel have had an enormous practical impact on many more individuals. The first two volumes in the Wind and Beyond series and the succeeding four now in preparation all cover the impact of aerodynamic development on the evolution of the airplane in America. As the six-volume series will ultimately demonstrate, just as the airplane is a defining technology of the twentieth century, aerodynamics has been the defining element of the airplane. The forthcoming volumes will proceed roughly in chronological order, covering such developments as the advent of commercial airliners, flying boats, rotary aircraft, supersonic flight, and hypersonic flight. This series is designed as an aeronautics companion to the Exploring the Unknown: Selected Documents in the History of the U.S. Civil Space Program (NASA SP-4407) series of books. As with Exploring the Unknown, the documents collected during this research project were assembled from diverse public and private sources. A major repository of primary source materials relative to the history of the civil space program is the NASA Historical Reference Collection in the NASA History Division. Historical materials housed at NASA Field Centers, academic institutions, and Presidential libraries were other sources of documents considered for inclusion, as were papers in the archives of private individuals and corporations. The format of this volume also is very similar to that of the Exploring the Unknown volumes. Each section in the present volume is introduced by an over- view essay that is intended to introduce and complement the documents in the section and to place them in a chronological and substantive context. Each essay contains references to the documents in the section it introduces, and many also contain references to documents in other sections of the collection. These introductory essays are the responsibility of Dr. Hansen, the series author and chief editor, and the views and conclusions contained therein do not necessarily represent the opinions of either Auburn University or NASA. The documents included in each section were chosen by Dr. Hansen’s project team from a much longer list initially assembled by the research staff. The contents of this volume emphasize primary documents, including long-out-of-print essays and articles as well as material from the private recollections of important actors in shaping aerodynamic thinking in the United States and abroad. Some key legisla- tion and policy statements are also included. As much as possible, the contents of these volumes comprise an integrated historical narrative, though Dr. Hansen’s

xxiii team encourages readers to supplement the account found herein with other sources that have already or will become available. Please note that the chapters in this series are numbered sequentially. Thus the first chapter in this second volume is referred to as chapter three and so forth. For the most part, the documents included in each section are arranged chrono- logically. Each document is assigned its own number in terms of the section in which it is placed. As a result, for example, the fifteenth document in the first chapter of this volume is designated “Document 3-15.” Each document is accompanied by a headnote setting out its context and providing a background narrative. These headnotes also provide specific information and explanatory notes about people and events discussed. Many of the documents, as is the case with Document 3-15, involve document “strings,” i.e., Document 3-15 (a-e). Such strings involve mul- tiple documents—in this case, five of them (a through e) that have been grouped together because they relate to one another in a significant way. Together, they work to tell one documentary “story.” The editorial method that has been adopted seeks to preserve, as much as possible, the spelling, grammar, and language usage as they appear in the original documents. We have sometimes changed punctuation to enhance readability. We have used the designation [abridged] to note where sections of a document have not been included in this publication, and we have avoided including words and phrases that had been deleted in the original document unless they contribute to an understanding of what was going on in the mind of the writer in making the record. Marginal notations on the original documents are inserted into the text of the documents in brackets, each clearly marked as a marginal comment. Page numbers in the original document are noted in brackets internal to the document text. Copies of all documents in their original form are available for research by any interested person at the NASA History Division or Auburn University. While the Exploring the Unknown series has been a good model in many ways, this volume indeed represents an expedition into uncharted waters. Dr. Hansen and his team have crafted a landmark work that will not only be an important reference work in the history of aeronautics, but interesting and informative reading as well. We hope you enjoy this useful book and the forthcoming volumes.

Dr. Steven J. Dick NASA Chief Historian Director, NASA History Division Acknowledgments

This volume represents the collected efforts of many members of an outstanding team. At Auburn University, a number of individuals provided generous assistance to Dr. James R. Hansen’s project team. Dr. Paul F. Parks, former University Provost, strongly encouraged and supported the project from its inception, as did Dr. Michael C. Moriarty, former Vice President for Research. To undertake his leadership of the project, Dr. Hansen gave up his job as Chair of the Department of History, something he would not have felt comfortable doing without being certain that the administration of his department would be in the capable hands of worthy successors—first, Dr. Larry Gerber, and then Dr. William F. Trimble. Both Gerber and Trimble gave hearty and vocal support to Auburn’s NASA history project. A number of colleagues in aerospace history gave help to the project, including Dis- tinguished University Professor Dr. W. David Lewis and Dr. Stephen L. McFarland. Dr. Roy V. Houchin, who earned a Ph.D. under Hansen, lent aid and comfort to the project team from his vantage point inside the U.S. Air Force. A number of Hansen’s graduate students helped the project in various ways, notably Andrew Baird, and Kristen Starr, as did Dr. David Arnold, also of the USAF, and Dr. Amy E. Foster of the University of Central Florida, who earned their Ph.D.s in aerospace history during the time period when this project was being conducted. A number of people in the NASA History Division deserve credit. Jane Odom, Colin Fries, and John Hargenrader helped track down documents from our Histori- cal Reference Collection. Nadine Andreassen provided much valuable general assistance and helped with the distribution. Interns Rebecca Anderson, Giny Cheong, Jennifer Chu, and Caitlin Gallogly also helped out tremendously. Also at NASA Headquarters, Tony Springer in the Aeronautics Research Mis- sion Directorate served as an invaluable sounding board on technical aeronautics issues. Now at the National Air and Space Museum, former NASA Chief Historian Roger D. Launius is owed a special debt of gratitude for providing the initial impetus and guidance for this worthy project. A talented group of professionals handled the production of this book. Heidi Pongratz at Maryland Composition oversaw the copyediting of this book. Tom Powers and Stanley Artis at NASA Headquarters acted as invaluable coordinating liaisons with the graphic design group at Stennis Space Center. At Stennis, Angela Lane handled the layout with skill and grace, Danny Nowlin did an expert job proofreading, and Sheilah Ware oversaw the production process. Headquarters printing specialist David Dixon expertly handled this last and crucial stage of production.

xxv Introduction to Volume II

Reinventing the Airplane

The history of aeronautical technology concerns much more than just the nuts and bolts of airplanes and spacecraft, much more than just the history of propellers and wings, more than the history of landing gear and jet engines, more than the ornithology of P-51s and F-22s, or the genealogy of X-planes. The history of flight technology is just as much a story of people and ideas as are histories dealing with any other topic related to society and culture. Scholars who write about the history of aerospace technology have plenty to say about the research, design, building, maintaining, and utilizing of flight vehicles, but their studies are no less human, no less connected to social, political, or cultural forces because they deal with technical matters. The history of aeronautical technology tells us a lot about our existence as a thinking, dreaming, planning, scheming, aspiring, and playful species. As aerospace industry analysts William D. Siuru and John D. Busick noted in relation to their study of the evolution of modern aircraft technology, humankind’s journey through the ages has been eased and accelerated, but also complicated by our unique and irre- pressible knack for technology and invention.1 From the stone ax and clay pot to the electron microscope and Human Genome Project, our technological creations have been ingenious, phenomenal, and occasionally—for good and for ill—of world- shaking significance. This is, by all means, true for the airplane, one of the most ingenious and phenomenal—if slow-to-come—inventions in history, and surely one of the most world-shaking. In how many ways has the flying machine changed society? As Antoine de Saint Exupéry wrote in 1939, it has “unveiled for us the true face of the Earth.”2 It has brought people together, changed our economy, added an unprecedented new dimension to warfare, affected everything from government, public administration, international relations, international policies, manufacturing, mar- keting, mining, cities, and real estate, to media, railroads, ocean shipping, agricul- ture, and forestry. It has affected population, the family, religion, health, recreation, education, crime, and even sex.3

1 William D. Siuru, John D. Busick, Future Flight: The Next Generation of Aircraft Technology, 2nd ed. (Tab Books, 1994), p. 3. 2 Antoine de Saint Exupéry, Wind, Sand, and Stars, translated from French by Lewis Galatiere (New York, 1939), p. 97. 3 One of the most remarkable analyses of the overall impact of aviation on the world came right after World War II with William Fielding Ogburn’s The Social Effects of Aviation(Boston, 1946).

xxvii xxviii The Wind and Beyond: Volume II Introduction to Volume II: Reinventing the Airplane xxix

It has not been all for the good. In the 90 years from the tragic death of Lt. results, the utopian hopes versus the spotted actuality, the what-might-have-been Thomas Selfridge in Orville Wright’s airplane at Fort Myers, Virginia, in 1908, against what actually happened, and the trade-offs among various ‘goods’ and pos- to the use of commercial airliners by terrorists to attack New York and Washing- sible ‘bads.’” All of these comparisons can be made “by seeing how technology ton, DC in September 2001, there has never been a time when aviation did not interacts in different ways with different values and institutions, indeed, with the know terrible accidents. Aviation has also brought human conflict to new depths of entire sociocultural milieu.”6 destruction. Despite this fact, the flying machine has always inspired “great expecta- But Kranzberg’s first law is not the only “law” apropos to consideration of the tions”—perhaps too great, given that it is, after all, just one of our many machines. history of flight technology. Another basic insight comes not from historians, but Orville Wright summed up our loftiest ambitions for aviation when he said that it from those who work in the aerospace industry. There is a saying in that industry: had been his hope (and that of brother Wilbur) that they were giving the world “an “Requirements push and technology pulls.” What this means, in a nutshell, is that invention which would make further wars practically impossible.” Unfortunately, the requirements of new missions, or even the need to improve upon current jobs, history proved them wrong, and it did not take long to do it. As much as we admire often drives engineers and scientists to work on the leading edge of technology. the “Bishop’s Boys” for their dream of a benevolent instrument of global peace, we They are being “pushed” by ever more demanding requirements to find solutions to are equally astonished by how such extraordinarily clear and logical thinkers could problems through the invention of new ideas. Technology then “pulls” by attracting have been so ordinarily naive about the forces in the world around them. Maybe those responsible for finding a way to meet the requirements for the newest con- someday their vision will be proved right, and the world will discover, as the Wrights cepts germinating in university, government, and commercial laboratories. For the did, that peace, like flight, requires not brute power, but control and balance. push and pull to work together effectively, it takes forward-thinking planning smart Contrary to what many engineers, most technocrats, and the great majority of enough to envision a way to use the new technology successfully in the design of a industrial entrepreneurs seem to believe; contrary to people who use the Internet brand new aircraft. to read the morning paper, or to golfers who cannot enjoy a round of golf without This sequence of developments—(1) requirements [or needs], (2) technology, riding in an electric golf cart and swinging a $500 titanium-headed driver; contrary and (3) concepts—has been, and still is, basic to the technological progress of most to what many people in modern consumer society seem to believe, technology is not modern aircraft—and perhaps all military aircraft. “Requirements push and tech- inherently good. In the words of one of the founding fathers of the history of technol- nology pulls” may be just a more complicated way of the old saying, “Necessity is ogy as a discipline, Melvin C. Kranzberg, “technology is neither good, nor bad, nor the mother of invention.” There is considerable common sense, and historical valid- is it neutral.” Kranzberg called this “The First Law of the History of Technology.”4 ity, to this aphorism, but it is also true that it is not always the case—or always that By its very nature, no technology is absolutely “good”—and none is bad. But illuminating of what actually is going on. Sometimes “necessity is not the mother neither is technology ever neutral. Depending on how we design technology, and of invention,” but rather “invention is the mother of necessity.” This was, in fact, even more on how we use technology, it will affect us, and change us, in some way. Kranzberg’s second law of the history of technology—and it makes us think about Whether the effects and changes turn out to be good or bad, or both inseparably aerospace technology in some very important ways. together, is not predestined in the inherent qualities of the technology itself but Once the Wrights invented the airplane, all sorts of things really needed to rather depends on the broader context and values within which we live our lives. happen. Over the course of the next 30 years, as this volume shows, the airplane The human consequences of the airplane have gone far beyond what the Wrights was in a sense reinvented as the Wrights’ achievement was completely rethought and or anyone else imagined in 1903. If it had been invented at a different time, or if it reworked by emerging groups of professionals dedicated to the airplane’s improve- had been introduced into a different context or under different circumstances, the ment and greater practicality. What Kranzberg’s second law illuminates is that invention of the airplane might have led to quite different results. In this case, as in “Every technical innovation seems to require additional technical advances in order others, “The river of history could have cut a different canyon.”5 to make it fully effective.”7In the case of the airplane, the invention quickly neces- Kranzberg’s first law reminds us to “compare short-term versus long-term sitated all sorts of auxiliary technologies: advanced structures and materials, new wing shapes, streamlined aerodynamics, retractable landing gear, efficient low-drag engine cowlings, variable-pitch propellers, and much more. But perhaps even more 4 Melvin C. Kranzberg’s classic essay “Technology and History: ‘Kranzberg’s Laws’” (in Technology and Culture 27 [July 1986]: 544-560) offers penetrating and witty analysis of the interactions between importantly, it also necessitated new social forms and organizations (e.g., military technology and its social context. 5 See George Will, “What Paths Would the Nation Have Taken Had Taylor Lived?” Washington Post, 6 20 June 1991. Will’s article raises fascinating issues relevant to the historical “what-ifs” and “might- “Kranzberg’s Laws,” 547-548. have-beens.” 7 “Kranzberg’s Laws,” 548-549. xxx The Wind and Beyond: Volume II Introduction to Volume II: Reinventing the Airplane xxxi

air services, airlines, airports, government bureaus, research laboratories, engineer- Let us hope that the flying machine, in the 21st century, does “free” us, in ing curricula, and much more) in order to make the airplane more fully practi- more positive ways, than it has been able to do in the century just passed. There is cable. “While it might be said that each of these other developments occurred in a no guarantee that it will. But like our dear Wright brothers gazing into their future response to a specific need,” Kranzberg claimed, “it was the original invention that that is our present, let us proceed into this new millennium with optimism that our mothered the necessity.” globe’s political environment will improve so that our future generations can enjoy It is important to underscore one last, essential point before moving into this our technical advances and not be destroyed by them. It is something in which the second volume of The Wind and Beyond. Just because the history of technology Wrights would want us not only to apply our best problem-solving and inventive involves technology, it does not mean that technical factors always take precedence. skills, but also in which to invest our limitless capacity to hope and to trust. In the real world, “soft” and “mushy” things such as politics and culture, for example, what bankers think can make them money or what activists say may harm the environment, often override good technical or engineering logic. And they should. Some might say that is why an American SST has never flown. That is why in the history of the American space program, all the thoughtful and well-intentioned talk about “the next logical step” has almost never led to it. After launching a man into space via Project Mercury, NASA said that the next logical step was to establish a permanent manned presence in low earth orbit, but instead the country landed men on the moon. After going to the moon via Project Apollo, the next logical step was to build an earth-orbiting space station along with a space shuttle to service it, but instead, the Nixon Administration decided that the country could not afford both and could manage temporarily with just the shuttle, although the space station had always been the shuttle’s main reason for existing. After the shuttle, surely the next logical step was to build a space station, but once again, the country found reasons to postpone building one. Clearly, logic does not determine the history of technology, and technologically “sweet” solutions do not always triumph over political and social forces. Historical logic, if we even want to use that phrase, is not the logic of engineers and scientists; it is the logic of Lewis Carroll’s Through the Looking Glass. In that all-too-real fantasy land, Tweedledee explains logic to Alice: “Contrariwise, if it was so, it might be; and if it were so, it would be; but as it isn’t, it ain’t. That’s logic.” Tweedledee’s logic is the only kind the American space program has ever known, or probably ever will. So, when stumbling across a book about the history of aerospace technology, a reader should not be put off because he might think the book, and author’s brain, is simply full of engineering tables and equations. There is a lot of “soft and mushy stuff” there also. It is what makes our species human, an essential part of what makes us brilliant, and a large part of what drives us nuts. It is what makes the history of technology one of the most complex and fascinating subjects one can possibly study. There may be a bigger message here as well. In 1998, Microsoft’s Bill Gates said about the Wright brothers’ invention in a speech he gave at Time Magazine’s 75th anniversary celebration of the airplane that, “We have to understand that engineering breakthroughs are not just mechanical or scientific, they are liberating forces that can continually improve people’s lives.” Biographies of Volume II Contributors

James R. Hansen, Professor of History and Director of the Honors College at Auburn University, has written about aerospace history for the past 26 years. His newest book, First Man: The Life of Neil A. Armstrong (Simon & Schuster, 2005), offers insight into the life and times of the first man on the Moon, but also sheds new light on many of the aerospace events and personalities that shaped America in the second half of the 20th century. His two-volume study of NASA Langley Research Center—Engineer in Charge (NASA SP-4305, 1987) and Spaceflight Revo- lution (NASA SP-4308, 1995) earned significant critical acclaim. His other books include From the Ground Up (Smithsonian, 1988), Enchanted Rendezvous (NASA Monographs in Aerospace History #4, 1995), and The Bird Is On The Wing (Texas A&M University Press, 2003). Jeremy R. Kinney is a curator in the Aeronautics Division, National Air and Space Museum, Smithsonian Institution, and holds a Ph.D. in the history of technology from Auburn, with a specialization in aerospace history. Prior to joining NASM, he served as the American Historical Association NASA Fellow in Aero- space History. D. Bryan Taylor is a doctoral candidate in the history of technology at Auburn. He holds an M.A. in history from Brigham Young University. Molly F. Prickett holds a B.A. degree in English from Auburn University. J. Lawrence Lee holds B.M.E., M.A., and Ph.D. degrees from Auburn. He is an engineer-historian at the Historic American Engineering Record, a branch of the U.S. National Parks Service, and is former chair of the History and Heritage Com- mittee of the American Society of Mechanical Engineering.

xxxiii

Chapter Three The Design Revolution

Destination Document:

The DC-3 freed the airlines from complete dependency on government mail pay. It was the first airplane that could make money just by handling passengers. With previous aircraft, if you multiplied the number of seats by the fares being charged, you couldn’t break even—not even with a 100 percent full load. Economically, the DC-3 let us expand and develop new routes where there was no mail pay.

C. R. Smith, President of American Airlines, ca. 1938, quoted by Robert J. Ser- ling, Eagle: The Story of American Airlines (New York: St. Martins, 1985), p. 110.

On a Streamline to the DC-3

In 1935, the National Aeronautics Association (NAA) awarded its prestigious Collier Trophy to the Douglas Aircraft Company for its new family of twin-engine transport planes. The trophy, awarded annually, recognized the greatest achievement in American aviation. In doing so, the NAA acknowledged not only a super- lative aircraft but also a revolution in the design of airplanes. In the three decades since the achievement of flight by the Wright brothers, the flying machine had evolved from a fragile contraption of wood, wire, and cloth into a sleek and sturdy

The Douglas DC-3 became the most popular and reliable propeller-driven airliner in aviation history and its appearance in the mid-1930s marked the culmination of the design revolution in American aircraft. (SI Negative No. ACC 1997-0033)

1 2 Chapter 3: The Design Revolution On a Streamline to the DC-3 3

form of modern public transportation. The Douglas transport honored by the 1935 problems. Solutions to these problems in the interwar period led in fundamental Collier Trophy featured all-metal construction, an enclosed streamlined shape with ways to the design and operation of advanced aircraft like the Douglas DC-3, which cantilever monoplane wings, cowled radial engines, controllable pitch propellers, were safer, faster, higher-flying, and generally more versatile and dependable than high-lift flaps, and retractable landing gear. One of the great airplanes of all time, any aircraft ever flown. New research and development tools, especially the wind the Douglas DC-3, represented not only what was to become one of the truly classic tunnel, played a key role in the process of reinventing the airplane, as engineers designs in aviation history but also the culmination of a creative process involving worked to create the most effective streamlined shape possible. an almost total “reinvention” of the airplane. One of the central problems addressed by the airplane design revolution of The reinvention of the airplane flowed from many technological streams that the 1920s and 1930s was aerodynamic drag reduction. In his history of aerody- converged in the years following World War I. Aviation technology had progressed namics, John D. Anderson has asserted that, “The major thrust in the age of the in significant ways in the fifteen years after Kitty Hawk, but strut-and-wire-braced advanced propeller-driven airplane can be summarized in a word: streamlining.1 Air- open-cockpit biplanes with fixed landing gear still pretty much reflected the state craft designers of the interwar period understood streamlining, in Anderson’s words, of the art. Aircraft designers worked from a very limited understanding of aerody- as “adopting a form that is so shaped and so free of protuberances that it produces namics, and their products were primarily the result of empirical, even “cut-and- no eddies in the airflow over it.” In essence, streamlining meant drag reduction, try,” engineering. A few potential breakthroughs appeared during the war, but frail and drag reduction enabled aircraft to fly farther and carry heavier loads. Reduced wooden biplanes covered with fabric, braced by wires, powered by heavy water- drag would also allow aircraft to fly faster without expending more energy and fuel cooled engines, and driven by hand-carved wooden propellers still ruled the airways resources, meaning that the same size airplane could perform the same work with in 1918. The principles of aeronautical engineering had yet to be fully discovered, less energy cost. All the way back to the days of Sir George Cayley, aeronautical and only a few programs at major schools such as Massachusetts Institute of Tech- experts had been aware of the presence of aerodynamic drag (i.e., the resistance nology (MIT) and the University of Michigan existed to find these ideas and teach to the forward movement of body in a fluid like air). But because the sources and them to students. Aircraft design remained a largely intuitive practice requiring even the actual composition of aerodynamic bold speculation and daring, in both a financial and technological sense. drag were not well understood, few quantita- In terms of engineering, a number of unknowns endangered aircraft perfor- tive or otherwise nonintuitive methods were mance. Even those few professional aeronautical engineers who did exist did not available to pioneering designers to lessen the know for sure how to reduce engine drag without degrading cooling. They did not effects of drag on aircraft performance. know with certainty how to shape wings to increase lift or diminish the effects of As with most revolutions, the aircraft turbulence. They did not know how and when flaps, ailerons, and other control design revolution of the 1920s and 1930s drew surfaces worked best. In addition, they did not even know if it was worthwhile to upon earlier ideas not materialized in tangible retract landing gears—according to certain pundits, the added weight and structural ways during their own day. Without question, complexity of a retractable undercarriage was not worth the saving in air resistance. the reinvention of the airplane depended on Aeronautical engineers suspected that substantial increases in aerodynamic efficiency this sort of heritage. It drew inspiration as well might follow on the heels of correct answers to just a few of these technical concerns, as practical lessons from what pioneers in the- but they did not know exactly how, or even whether to try, to get at them. oretical aerodynamics discovered back in the In significant ways, the process of “reinventing the airplane” that unfolded in late 19th century. the 1920s and 1930s grew naturally out of these questions and suspicions. Institu- In key respects, it was the work of English tionally, the process took shape and gained momentum as a growing body of profes- engineer and automobile maker Frederick W. sionals began to attack the problems obstructing the immediate progress of aviation, Lanchester that inaugurated the first great age particularly those vexing the fledgling military air services and aircraft manufactur- of aerodynamic theory. Curiously, Lanchester’s Frederick W. Lanchester’s experimental and nonquantitative search for a “streamlined ing and operating industries. Major new institutions played critical roles. In the work was largely experimental and nonquan- form” resulted in his 1907 pioneering book, United States, the National Advisory Committee for Aeronautics (NACA) and the titative, involving a program of experiments Aerodynamics. (SI Negative No. 45374-C) research agenda of its Langley Memorial Aeronautical Laboratory was formative in defining what needed to be accomplished. Following its establishment in 1915, the NACA conducted research into basic aerodynamic, structural, and propulsion 1 John D. Anderson, Jr., A History of Aerodynamics and Its Impact on Flying Machines (Cambridge University Press, 1997), p. 319. 4 Chapter 3: The Design Revolution On a Streamline to the DC-3 5

initiated in the early 1890s, with cambered airfoils. The principle objective of Lanches- ter’s work was to define an efficient low-drag shape with smooth contours that offered no resistance to the passage of airwaves along its surface; he called this shape a “streamline form.” Lanchester realized that irregularity in an airfoil’s shape caused disturbances in an airflow and resulted in a significant amount of drag. Unfortunately, Lanchester’s work did not receive the attention it deserved at the time, in part because his theories lacked the type of truly quantitative methodology that could have practically aided early aircraft designers. Document 3-1 provides excerpts from Lanchester’s 1907 book, Aerodynamics, featuring his concept of streamlining. German physics professor Ludwig Prandtl Another formative aerodynamic thinker pioneered the academic study of drag and the for the subsequent revolution in airplane use of streamlining to improve aerodynamic design of the 1920s and 1930s was the influen- efficiency overall in the 1920s and 1930s. (SI Grover Loening (left), seen here with Orville Wright (right) in 1913, was one of the earliest aeronautical engineers Negative No. 74-10601) tial German physics professor Ludwig Prandtl, employed by the United States government and advocated the streamline form as the ideal shape for an airplane. much of whose work supported Lanchester’s (SI Negative No. 80-5407) ideas about streamlining. The best understanding of aerodynamic drag at the turn of the century was based on what little was known about the performance of hydro- he taught about airflow in the boundary layer near the surface of an aerodynamic dynamic forms. Yet both Lanchester and Prandtl realized that air and water were far body provided the first rational basis for calculating what came to be known as different media—for one example, air created more frictional resistance than water. “skin friction drag.” His theory also promoted vastly improved understanding of Addressing this issue at the Third International Congress of Mathematicians at Hei- the dynamics of flow separation around an aerodynamic body, which was a major delberg in 1904, Prandtl observed that an ideal airflow moved in smooth parallel source of “ form drag.” Prandtl’s students also learned that the best way to lessen layers. But in real-world circumstances, such as the flow of air over a wing surface, the effects of “parasitic drag” (i.e., the combination of skin friction drag and form this so-called laminar flow became turbulent, resulting in significant drag. What drag) was to delineate smoother contours for wings and other aerodynamic forms. Prandtl reported in 1904—to a largely disinterested audience of purer-minded In sum, Prandtl’s work, like Lanchester’s, not only highlighted the problem of drag, mathematicians—was that an airflow changed in dramatic fashion from zero to it underscored the practical advantages of streamlining. constant velocity in a very thin “transition layer” right next to the surface of the air- While the work of Lanchester and Prandtl provided an essential theoretical foil. In this extremely narrow zone, later to become known as the “boundary layer,” basis for understanding drag, the concept of streamlining saw little practical appli- transition from laminar to turbulent flow affected aircraft performance profoundly.2, cation in the vast majority of aircraft designs emerging from World War I. Angular Prandtl’s boundary-layer hypothesis of 1904 (excerpted as Document 3-2) offered shapes and drag-producing protuberances were still ubiquitous features. Neverthe- major new opportunities for aircraft designers and aeronautical engineers. What less, the idea of improving aircraft performance through streamlining did catch the attention of some people in the aeronautical community during the war. One such individual was Alexander Klemin, a civilian research working for the U.S. Army Air 2 The first known reference to the term “boundary layer,” according to the Oxford English Dictionary, came in 1921, when Dutch fluid mechanics specialist J. M. Burgers wrote in his country’s Proceedings Service, who explored the idea of streamlining in a 1918 textbook entitled Aeronau- of the Royal Academy of Science (Proc. K. Acad. Wetensch [Amsterdam]) that “we can calculate tical Engineering and Airplane Design, one of the first works of its kind published in the distribution of the vorticity in the boundary layer, when we suppose the velocity outside the the United States (see Document 3-3). A graduate of the aeronautical engineering boundary later to be known.” The second recorded use of the term, according to theOED , came in an article appearing in Flight magazine on 20 November 1924: “The deductions from the boundary program at MIT, Klemin came to head the Aeronautical Research Department at layer theory gave a rather poor approximation to the truth.” the U.S. Army Air Service’s engineering facility at McCook Field, in Dayton, Ohio. 6 Chapter 3: The Design Revolution On a Streamline to the DC-3 7

efficiently extending its range. To achieve an economic range through the improvement of the lift-to-drag ratio, Breguet called for the use of streamlining, specifically suggesting such measures as the use of retractable landing gear. He even defined a target “fineness” ratio for commercial aircraft—a ratio ultimately achieved just over a decade later in the design of the Douglas DC series. Another factor that proved to play a critical role in spurring both the development and general acceptance of streamlined aircraft designs was performance at air races, which became enormously popular in Europe and America in the late 1920s and 1930s. At the international level, the greatest of these races, such as the Schneider Trophy competition, presented opportunities for benign competition between world powers; victories won and records set were perceived by all as signs of national technological prowess. Within a given country, races held similar significant meanings; for example, in the United States, the Army and Navy worked feverishly to come in first in the competitions, hoping through first-place finishes and world and national records to secure greater political support for the development of the air forces. The air races fostered experimentation and Aircraft such as the biplane, strut-and-wire braced Martin MB-1 Bomber represented the state of the art in rapid technological development, much as auto- aerodynamic design circa 1917. The designers of the aircraft, Laurence D. Bell (left), Glenn A. Martin (second mobile races did and still do. Both the Army and from right), Donald W. Douglas (right), and James Kindleberger (not pictured), would use the advances of the design revolution to create vastly different aircraft. (Martin test pilot, Thomas E. Springer, is second from left.) Navy used wind tunnels to refine their racers to (SI Negative No. 43067) get the utmost speed out of them. Many racing French aircraft builder Louis Breguet made aircraft employed groundbreaking devices that one of the strongest cases for the social would become part of the modern airplane; and economic benefits of aerodynamic Equating “resistance” to “drag,” he noted in his groundbreaking textbook that a streamlining in a 1922 speech delivered streamline form was the ideal shape for an airplane. But major obstacles blocked these included streamlined designs, cantilever before the Royal Aeronautical Society en- the achievement of ideal streamlining, notably the realities of conventional design wings, and retractable landing gear. In 1923, an titled, “Aerodynamical Efficiency and the American airplane, the U.S. Navy Curtiss CR3s, Reduction of Transport Costs.” (SI Nega- practice, which necessitated the incorporation of mammoth and strangely shaped tive No. 77-542 or 78-13907) engines, the bodies of pilots and passengers, plus messy structural items such as took first and second place in the Schneider Cup bracing wires and struts. Somehow, Klemin concluded, many of these obstacles international seaplane race in Cowes, England, the winner at a top speed of 177.38 would have to be overcome if an airplane ’s aerodynamic effectiveness were ever to miles per hour (mph). The aircraft employed new technology, including flush wing improve much. radiators and a solid aluminum alloy propeller. This propeller incorporated aero- After the war, more and more aviation advocates placed an emphasis on air- dynamically advanced thin airfoil sections based on the designs of Dr. Sylvanus A. craft efficiency, culminating in 1922 when pioneering French aircraft manufacturer Reed. At the tips of the propeller blades, the airflow reached speeds approaching the Louis Breguet issued a clarion call for the development of streamlined designs. In speed of sound. But it was the sleek, streamlined fuselage of the aircraft that really an address before the Royal Aeronautical Society in London entitled “Aerodynamic captured everyone’s eye. As contemporary observers noted, everything about the air- Efficiency and the Reduction of Aircraft Costs” (see Document 3-4), Breguet out- plane spelled “speed” (see Document 3-5). Another Navy Curtiss racer, the R2C-1, lined the problem in both technical and economic terms, and he articulated vari- won the 1923 Pulitzer Trophy Race in St. Louis, Missouri. In doing so, the slender ous social incentives for the improvement of civilian aircraft in the postwar period. little airplane set new world speed records of 243.8 mph for 100 km and 243.7 mph Breguet believed that an improvement in the lift-to-drag ratio, which he called the for 200 km over a closed circuit. No other aircraft flying at the time could boast “fineness” of an aircraft design, would enhance an airplane’s economic potential by cleaner aerodynamics. Curtiss designers got the airplane’s zero-lift drag coefficient 8 Chapter 3: The Design Revolution On a Streamline to the DC-3 9

The challenges and glories of air racing definitely inspired streamlining, but in the minds of many, the association of streamlined shapes with racing implied that the technology of highly sleek and efficient aerodynamics belonged only to the realm of high-speed applications. But this was hardly the case. Advanced aerodynamic refinement in the form of streamlining was already beginning to play a critical role in the design of a number of revolutionary civil and commercial aircraft as well, a technological development that would help stimulate the first great age of aviation as an effective mode of mass public transportation. One of the first nonmilitary or racing aircraft to demonstrate streamline design was the boldly innovative Lockheed Vega of 1926, designed by engineering genius John K. Northrop. Self-taught, highly creative, and remarkably proficient in transferring a speculative design from his “mind’s eye” to reality, “Jack” Northrop stood ready at the leading edge of the streamlining movement in the United States, albeit from a direction quite different from that of building racers or being immersed academically in aerodynamic The streamlined fuselage, flush wing radiators, and solid aluminum alloy Reed propeller of the navy Curtiss CR-3 racer highlighted the role of aerodynamics in high-speed aircraft design. Lt. David Rittenhouse (in cockpit) flew theory. His “feel” for aerodynamic refinement the CR-3 to victory in the 1923 Schneider Cup international seaplane race in Cowes, England. (SI Negative No. was more “aesthetic,” in that it involved a A-47217) heightened sensitivity for what looked beautiful and was in “good taste.” Not that Northrop (C )—considered by many experts as the best indicator of the aerodynamic D,O did not base his designs in engineering prac- cleanness or refinement of an aircraft.3 In fact, it ranks among the lowest values in ticality. The key to the Vega’s aerodynamic the entire history of aerodynamics for propeller-driven aircraft—lower than that efficiency rested in the techniques he used enjoyed by four of the sleekest aircraft produced by the airplane design revolution of Jack Northrop’s innovative construction tech- in its construction. Lightweight, strong, and the 1930s: the Lockheed Vega (0.0278), Lockheed Orion (0.0210), Boeing 247D niques facilitated his design of highly efficient remarkably streamlined, the Vega represented and aesthetic streamline aircraft designs such (0.0212), and Douglas DC-3 (0.0249). Only a few American prop-driven planes a dramatic departure from previous designs in as the Lockheed Vega and the Northrop Alpha. have ever achieved lower zero-lift drag coefficient. Noteworthy among these were (SI Negative No. 75-5442) that it was a single-engine, high-wing cantile- the Beechcraft D17S four-place monoplane of 1939 (0.0182), North American ver monoplane with a semimonocoque (monocoque meaning “one shell”) plywood P-51D fighter plane of 1944 (0.0163), and Beech Bonanza V-35 general aviation stressed skin fuselage. The cantilever wing featured internal bracing. A major new aircraft of 1970 (0.0192). feature incorporated in later Vega airplanes was a circular cowling surrounding the 450-horsepower Pratt & Whitney Wasp air-cooled engine. This cowling concept 3 For a clear explanation of the significance of the zero-lift drag coefficient, seeL aurence K. Loftin, Jr., derived from the NACA’s systematic identification of efficient cowling forms, ones Quest for Performance: The Evolution of Modern Aircraft (Washington, DC: NASA SP-468, 1985), that not only reduced drag but also at the same time, improved cooling of the engine. pp. 4-5, 158-160. Loftin defined this coefficient Cd, o as “a nondimensional number that relates The only major nonstreamline element of the Vega’s profile was its fixed landing gear, the zero-lift drag of the aircraft, in pounds, to its size and the speed and altitude at which it is flying. Generally speaking, the smaller the value of the number, the more aerodynamically clean but in later versions, even that had fairings called “pants” around its wheels, which the aircraft.” As useful as it is as a measure of aerodynamic refinement, the significance of the zero- also helped to reduce drag. As mentioned previously, the airplane benefited from a lift drag coefficient is limited in application because it is based on wing area and because, for a very low zero-lift drag coefficient, which helped it to reach a maximum speed of 190 given wing area, several different sizes of fuselage and tail may be employed. Thus, as Loftin made clear, “differences in zero-lift drag coefficients may be interpreted as a difference in aerodynamic mph. Airlines used the plane to fly passengers (six at a time), and a number of pilots, refinement” when, in fact, the difference may be the product of differences in the ratio of “wetted” including Amelia Earhart and Ruth Nichols, broke records in it. In a Vega named area (i.e., the area of the entire body that comes in direct contact with the airflow) to wing area Winnie Mae, Wiley Post flew solo around the world in the summer of 1933, need- (Loftin, pp. 158-159). 10 Chapter 3: The Design Revolution On a Streamline to the DC-3 11

observed that the leading developer of many of these “shelf items” was the NACA itself. In his view, “no other institution in the world contributed more to the definition of the modern airplane”.5 The improved components, along with the fundamental understanding of the mechanism of drag that resulted from NACA research, became widely disseminated throughout the aeronautical engineering community and the aircraft industry during this crucial transition period. Without question, the NACA contributed greatly to virtually every aspect of airplane aerodynamics during the late 1920s and early 1930s. Some of the most important work was in the development of improved airfoils, a subject treated in detail in the next chapter. But the NACA team at Langley laboratory in Virginia also worked on many other areas of aircraft component improvement. For example, some of the first projects in the new Propeller Research Tunnel (PRT), which began operations in July 1927, concerned improvement of the aerodynamic qualities of engine installations, struts, and landing gear. The first important step in this work was to determine through systematic experimental parameter variation exactly how The Lockheed Vega of 1926 represented a dramatic departure from previous designs with its monocoque construction, NACA cowling, and teardrop wheel pants. Wiley Post flew the Lockheed Vega 5-B “Winnie Mae” solo much drag a particular component produced. around the world in only seven and one-half days during the summer of 1933. (SI Negative No. A-47516) The test program that initiated the NACA’s move toward the widespread reduction of drag focused on engine cowlings. Aerodynamic tests on one of the Army’s ing only seven and one-half days. In Document 3-6, Northrop recorded some of his small Sperry Messenger airplanes in Langley’s PRT in late 1927 revealed that the views on streamlining and its relationship to construction technology as embodied exposed cylinders of the air-cooled radial engine caused 17 percent of all drag plagu- in the Lockheed Vega. ing the aircraft. The most obvious solution to this problem was simply to cover the The development of streamlined aircraft did not come all at once. Perhaps more engine with some sort of streamlined shroud or cowling. But such a covering, it was than any other single element, it awaited widespread recognition of the advantages thought, would restrict airflow past the cylinders and cause the engine to overheat. of streamlining by some of its most influential leaders, if not the aeronautical com- With this in mind, cowlings were not used very often. But the drag problem grew munity as a whole (see Document 3-7). The initial focus of the efforts to reinvent into a more and more critical concern, so much so that at the NACA’s first annual the airplane hinged on the independent development of specific airplane parts, what manufacturers’ conference, held at Langley in May 1926, representatives of the air- aviation historian Richard K. Smith referred to as “shelf items.4 These included craft industry and the U.S. Navy’s Bureau of Aeronautics identified it as a national components such as airfoils, engine cowlings and nacelles, flaps, propellers, fillets, priority. The challenge for the NACA was to define a form of cowling that signifi- and retractable landing gear. Government researchers, engineers in industry, and cantly improved aerodynamic efficiency without degrading cooling.6 lone inventors played a part in developing these innovations—for the most part, Documents 3-8, 3-9, and 3-10 all relate to the NACA’s cowling program, for separately and individually. Designers and manufacturers would then incorporate which the NACA won its first Collier Trophy in 1929. The first of these documents and synthesize the new technology into final aircraft design as needed. Despite their is from the autobiography of Fred E. Weick, the Langley engineer who master- “shelf-item” status, these innovations were not truly “stand-alone” technologies that minded the development of the NACA’s low-drag cowling. In it, he recalled in could simply be put on the airplane; they had to be integrated into a total design. One important function performed by the NACA was full-scale wind tunnel testing of the aerodynamic advantages inherent to all these various components as they were 5 Smith, “Better,” p. 240. located on new airplanes, which provided data tables and computations so designers 6 For a complete analysis of the history of the NACA cowling program, see James R. Hansen, could use the information for a variety of future designs. Richard K. Smith further “Engineering Science and the Development of the NACA Low-Drag Engine Cowling,” in From Engineering Science to Big Science: The NACA and NASA Collier Trophy Research Project Winners, ed., Pamela E. Mack (Washington DC: NASA SP-4219, 1998), pp. 1-27. This chapter is an updated and expanded version of Chapter 5 of Hansen’s Engineer in Charge: A History of the Langley Aeronautical 4 Richard K. Smith, “Better: The Quest for Excellence,” in Milestones of Aviation, ed. John T. Laboratory, 1917-1958 (Washington DC: NASA SP-4305, 1987), entitled “The Cowling Program: Greenwood (New York: MacMillan, 1989), p. 240, 243-4. Experimental Impasse and Beyond,” pp. 123-139. 12 Chapter 3: The Design Revolution On a Streamline to the DC-3 13

The inclusion of the NACA low-drag cowling to the Texaco Lockheed Air Express allowed Frank Hawks to establish a new Los Angeles to New York nonstop record of 18 hours, 13 minutes in February 1929. (SI Negative No. The inclusion of NACA Cowling No. 10, seen here undergoing tests in the 20-foot tunnel in September 1928, A31250-G; Videodisc No. 2B-11966) into the design of the Martin B-10 bomber increased the airplane’s maximum speed by 30 mph to 225 mph and reduced its landing speed significantly. (NASA Image No. 1974-L-02730)

Hubert C. H. Townend of the British National Physical Laboratory developed a low-drag engine cowling in 1929. Known as the “Townend ring,” the design reduced drag with little degradation of engine cooling and was widely adopted in Europe and America. (SI Negative No. 76-17366)

The NACA won its first Collier Trophy in 1929 for its innovative work on low-drag cowlings. (NASA Image No. L-1990-04348) 14 Chapter 3: The Design Revolution On a Streamline to the DC-3 15

extraordinary detail the careful experimental process by which the NACA arrived at its award-winning shape. The second involves a string of documents, the first of which relates to the stunning performance in February 1929 of a Lockheed Air Express, a derivative of the Lockheed Vega. Equipped with a NACA low-drag cowling that increased its speed from 157 to 177 mph, this aircraft, piloted by Frank Hawks, established a new Los Angeles to New York nonstop record (18 hours and 13 minutes). The third presents a trio of documents concerning the appearance of other new forms of engine cowlings that came to rival the NACA design. During the process of developing better components for an aircraft, the work on one “shelf item” often led to study of different problems. This situation happened in regard to the NACA cowling, as research engineers found that the cowl did not benefit all aircraft equally: its effectiveness depended on the shape of the airplane behind it. While these complexities spurred on additional work related to cowlings, it also led to other lines of investigation. The original purpose of NACA’s Langley Propeller Research Tunnel (PRT) was to develop improved propellers. But early tests in the PRT demonstrated that the machine could be used to study other problems, notably the overall effect of a propeller in combination with the placement of an engine and the placement of engine nacelles in relation to wings. These unintended lines of research led not only to data very significant to aircraft design- The alternative to the NACA cowling was the Townend ring, seen here mounted on a Boeing P-26 Peashooter perched in the Langley Full-Scale Tunnel for aerodynamic testing. (NASA Image No. L-09819) ers but to a few entirely new concepts. In studying the overall effect of the propeller in combination with the placement of the engine, for example, NACA researchers came to express their results in a new quantitative relationship they called “propulsive efficiency.” This term reflected the overall propeller output in reference to its mounting on the aircraft. Propulsive efficiency calculations soon became a standard analytical tool in the design of propeller-driven airplanes and played a crucial role in the development of modern streamlined aircraft. As Fred Weick recalled in our excerpt from his autobiography (Document 3- 8), another one of these inadvertent research programs, one devoted to the proper placement of engine nacelles, originated in 1929 when the PRT team at Langley began testing cowling forms on a multiengine aircraft, the Fokker trimotor. When speed trials with the big transport proved extremely disappointing, Weick and his associates started to wonder how the position of the nacelles with respect to the wing might be affecting drag. This was a critical design revolution, especially for multiengine aircraft, as big commercial and military aircraft were bound to be. In the case of the Fokker (as well as the Ford) trimotor, the original design location of the wing engines was slightly below the wing’s surface. As the air flowed back between the wing and nacelle, the expansion required was too great for the air to flow over the contour smoothly. The NACA Langley flight research division, in association with the PRT team, tried fairing in this space, but they achieved only a small improvement. Eventually the NACA’s systematic empirical approach yielded dividends. With the help of his assistants, Weick laid out a series of model tests in the PRT with The NACA worked to improve existing designs such as the Fokker Trimotor with the addition of experimental NACA-cowled nacelles placed in 21 different positions with respect to the wing: low-drag cowlings and streamline engine nacelles in 1929. (NASA Image No. L-03333) 16 Chapter 3: The Design Revolution On a Streamline to the DC-3 17

above it, below it, and within its leading edge. The resulting data on the nacelle’s It is perhaps curious to those unfamiliar with the history of aircraft technol- effect on the lift, drag, and “propulsive efficiency” of the big Fokker trimotor made ogy to learn that it took so long for wing flaps to become commonplace. The basic it clear that the optimum location of the nacelle was directly in line with the wing idea for them had been around since the aileron device invented by Frenchman and with the propeller fairly well ahead. Although their primary emphasis was on Henry Farman in 1908 to get around the Wright brothers patent for wing warping. drag and improved cooling, the tests at Langley also confirmed that NACA cowling Farman invented what came to be known as the aileron. Various inventors had cre- No. 10—the form that completely covered the engine but still managed to cool it by ated aileron devices prior to Farman, but they had all been separate rotatable sur- directing air to its hottest spots through ducts and baffles—could actually increase faces placed in front of (not behind) a wing or sometimes, in the cases of biplanes, the lift of the airplane’s wing in some cases, if the engine was situated in the opti- even between them. What distinguished Farman’s aileron was that the surface was mum position.7 integrated directly into the wing as aircraft designers still incorporate them today. In transmitting this important information confidentially to the Army, Navy, The modern definition of an aileron says that it is “a moveable control surface or and industry, the NACA helped build a several-months lead for American aircraft device, one of a pair or set located in or attached to the wings on both sides of an air- designers over rival European companies. After 1932, nearly all American transport plane, the primary usefulness of which is controlling the airplane laterally or in roll 8 and bombing airplanes—including the Martin B-10, Douglas DC-3, Boeing B-17, by creating unequal or opposing lifting forces on opposite sides of the airplane. But and many other famous aircraft of the era that followed—employed radial wing- this definition applies to the word aileron only after Farman’s innovation of 1908. mounted engines with the NACA-cowled nacelles located approximately in what A few ideas for flaps sprouted during World War I, but not actually that many, Weick and his associates had identified as the optimum position. Without question, since flying speeds were still too low to make such devices useful. As John D. Ander- this combination led to an entire new generation of highly effective airplanes with son explains in his history of aerodynamics, “the low wing loadings made flaps 9 which airlines for the first time could become financially self-supporting and no essentially redundant and most pilots rarely bothered to use them. After the war, longer in need of government subsidies. Paradoxically, as Document 3-11 suggests, engineers in various countries pursued new concepts for high-lift mechanisms. In some leaders of the U.S. aircraft industry at the time did not fully credit NACA a case of nearly simultaneous invention occurring just before 1920, Dr. Gustav V. research for this essential contribution to what was quickly becoming a blossoming Lachmann in Germany and Frederick Handley Page in England invented the “wing design revolution. slot.” Lachmann’s original design called for simply a long, spanwise slot located near Another critical component in the reinvention of the airplane was the develop- the leading edge. His idea was to create a pressure differential between the lower ment of high-lift devices, primarily flaps. Essentially a section of the wing’s trail- and upper surfaces of the wing. This differential would impel a high-energy jet of ing edge that could be hinged downward to increase the camber of the wing, a air through the slot. This jet flowed tangentially over the top surface of the wing, flap promised to boost lift and reduce the aircraft’s stalling speed. It permitted a along the way energizing the transition layer (i.e., boundary layer); delayed flow better angle of approach and lower speed on landing, important considerations not separation to much higher angles of attack; prevented stall; and increased lift in only for aircraft efficiency but also for safety. The earliest trailing edge flaps simply wind tunnel tests by a whopping 60 percent. Frederick Handley Page, who pooled increased aircraft drag during landing and approach instead of boosting lift, but the patent rights with Lachmann in 1921 (Lachmann actually went to work for Page’s desirable element was a structure that did both. In the 1930s, the introduction of company in 1929), went one step further. He combined a slot and a flap to create high-lift devices such as flaps allowed engineers to design aircraft with higher wing the slotted flap, a design that exposed a slot between the flap and the wing when loadings (i.e., the ratio of the gross weight of an airplane to the total planform area deflected. In combination with the thick new airfoil sections then being designed of its wing or wings). This was a new aerodynamic technology that proved crucial (see the next chapter), the Page slotted flap reliably produced even greater lift. Still, to the transition from biplanes to monoplanes. airplane designers mostly stayed away from flaps. In the United States, this neglect included a new type of “split flap” invented in 1920 by Orville Wright and associate J. M. H. Jacobs in a small laboratory provided by the Army at McCook Field in Dayton. In the 1920s, it was hard to find any aircraft with flaps except for those 7 We considered including the following NACA technical reports from 1932 in this chapter collection designed by Page and Lachmann. of documents, but, due to length, determined, it would suffice simply to reference them in a footnote: Donald H. Wood, “Tests of Nacelle-Propeller Combinations in Various Positions with Reference to Wings. Part I. Thick Wing—NACA Cowled Nacelle—Tractor Propeller,” NACA Technical Report

415 (Washington, 1932); and Donald H. Wood, “Tests of Nacelle-Propeller Combinations in 8 Various Positions with Reference to Wings. II.—Thick Wing—Various Radial-Engine Cowlings— Frank Davis Adams, Aeronautical Dictionary (Washington DC: NASA, 1959), p. 7. Tractor Propeller,” NACA Technical Report 436 (Washington, 1932). 9 Anderson, A History of Aerodynamics and Its Impact on Flying Machines, p. 365. 18 Chapter 3: The Design Revolution On a Streamline to the DC-3 19

It was not until the early 1930s that the idea of flaps started to entice many What NACA researchers most contrib- airplane designers—the result of the rapidly increasing speed of aircraft and their uted was systematic correlation between airfoil higher and higher wing loadings. The string of items comprising Document 3-12 theory, model propeller tests, and full-flight traces the development of one of the most attractive and ultimately successful of testing to find the best method of designing these American high-lift designs, known as the “Fowler flap.” Invented in 1924, this aerodynamically efficient propellers. Differ- was an extensible trailing-edge flap that increased both the camber and wing area. ences between the three sources of information Fowler flaps came to be used on several new aircraft from the 1930s on, notably about propellers showed that a significant gap planes built by Glenn L. Martin (in 1933, Martin hired Fowler to design flaps for still existed between empirical and theoretical him), as well as the Lockheed 14 twin-engine airliner of 1937. The Boeing B-29 knowledge. The first method, involving what bomber of World War II employed Fowler wing flaps (although they did not help was called “blade element theory,” assumed much to improve the high stalling speed of 105 mph for the big plane); so, too, did that a propeller blade was a series of isolated later versions of the Lockheed P-38 “Lightning,” for greater maneuverability. The airfoil sections that represented an ordinary wings of the Boeing B-17 employed very powerful Fowler flaps, designed specifically wing as they traveled in a helical path. Blade 10 to keep the landing speed within acceptable limits. The Germans even gave their William F. Durand’s long career in aeronautics element theory enabled engineers to design Messerschmitt ME 262 jet fighter Fowler high-lift flaps, along the trailing edge of included his pioneering, with Everett P. Les- propellers that were 70 to 80 percent efficient. ley, a standard table of aerodynamic design the wing (and in combination with full-span “slats”—that is, long narrow vanes or coefficients for propellers from 1917 to 1926. The multitalented Fred Weick, who authored auxiliary airfoils—in the leading edges). After the war, Boeing’s B-52 bomber aug- (NASM Videodisc No. 2B-57570) a textbook on propellers for McGraw-Hill in mented its lift via Fowler single-slotted flaps located at the trailing edge of the wing. 1931, said that blade-element theory enabled So Fowler’s invention of 1924 had a long shelf life. no more than a “cut-and-try process” when it came to designing new airfoils. Although the Fowler flap was only one of many different flaps designed and The second method involved testing model propellers in wind tunnels. The used after 1930, it is enlightening in Document 3-12 to follow how this one brand idea was to acquire reliable data from scale models that, through the law of simili- of flap disseminated through the American aeronautics community during the early tude, could be applied to similar geometric bodies of full size. By the 1920s, this 1930s. Similar studies could be made with the split flap, which was actually the type type of testing already had a distinguished history. From 1917 to 1926, professors first tried on American aircraft. The Northrop Gamma used split flaps, as did the William F. Durand and Everett P. Lesley of Stanford University conducted a pro- Douglas DC-1, in 1932, and the DC-3, in 1935. Split flaps ruled the roost early gram of “Experimental Research on Air Propellers” that provided a standard table on—some suggest it happened in deference to Orville Wright, one of its inventors. of propeller design coefficients, which any designer could utilize.11 Unfortunately, About the time designers started to convert to Fowler flaps and other sorts of slot- to make the data applicable for design, one had to carry out mathematical converted flaps, the double-slotted flap appeared—on Italy’s 1937 M-32 bomber and the sions that did not always work out right. It was thus vital to correlate experimental Douglas A-26 bomber of 1941. From the DC-6 on, Douglas put double-slotted results with theory and with full-scale testing if the resulting propeller shape was to flaps on the wings of all its airliners. The triple-slotted flap made its first appearance perform as expected. with the Boeing 727 jet airliner in the early 1960s. The final method, full-scale testing, represented the final, ultimate check for Perhaps an even more critical area of aerodynamic refinement necessary for the propeller efficiency and design. NACA researchers in the 1920s found that the airplane design revolution of the interwar period involved advancing the state of efficiencies of full-scale propellers were often six to ten percent greater than the the art in propeller performance. Again, the NACA played a fundamental role, con- efficiency of corresponding model propellers. A NACA report of 1925 attested that ducting and supporting propeller research on a consistent basis from its inception researchers “can never rely absolutely” upon model data until they verify it through as an organization in 1915. The committee’s first Annual Report acknowledged the full-scale flight tests. Testing the real thing in the actual flight environment, how- need “for more efficient air propellers, able to retain their efficiency over a variety ever, was expensive, time-consuming, potentially dangerous, and often inaccurate if of flight conditions.” In the ensuing years, one of its strongest and most regularly researchers used poorly maintained testing equipment or inadequate techniques. funded programs focused on the improvement of propellers.

11 See Walter G. Vincenti, Chap. 6, “Data for Design: The Air Propeller Tests of W. F. Durand and E. P. Lesley, 1916-1926,” What Engineers Know and How They Know It: Analytical Studies from 10 laurence K. Loftin. Jr., Quest for Performance, pp. 124, 132, and 139. Aeronautical History (Baltimore and London: Johns Hopkins University Press, 1990), pp. 136-169. 20 Chapter 3: The Design Revolution On a Streamline to the DC-3 21

Increased awareness of, and exposure to, the inadequacies of theory, wind tunnel experiment, and flight testing convinced the NACA that some form of “hybridiza- tion” of the three methods was necessary. In 1925, the NACA authorized the construction of the 20-foot PRT at Langley. In the following years, Fred Weick, Donald H. Wood, and other Langley researchers conducted various test programs that contributed to the fundamental development of the propeller—especially regarding the use of metal propellers. Their investigations found that by using blades that were thinner toward the tips, they could increase propeller efficiency. Propeller designers had already been aware that thin blade sections were ideal for high-speed applications because they did not suffer from what soon came to be known as “compressibility burble,” which amounted to a sharp increase in drag at high speeds. But until propellers were manufactured in metal, this benefit could not be realized. Wooden propellers and their thick sections, necessary for structural integrity, suffered severe drag limitations at high speeds. To turn faster, propellers needed to be thinner, which required the strength of metal. A significant breakthrough in propeller aerodynamics came in the form of the metal, multipiece, variable-pitch propeller—a technological development mostly associated with a research group, from the late 1910s into the late 1920s, under the leadership of Frank Caldwell in the Army’s aircraft engineering division at McCook Field. During the first quarter of the century, aircraft propellers had all come with fixed pitch, meaning that the angle at which each propeller blade struck the air remained static. Caldwell’s group at McCook recognized earlier that the ideal pitch Lindbergh gambled that a ground-adjustable pitch propeller set just below the optimum setting for cruise would of a propeller was not the same for different flight conditions. Low-pitch settings give him long-range efficiency while safely getting the Spirit of St. Louis off the ground at Roosevelt Field. The were more efficient for taking off and climbing, while high-pitch settings were better resultant dramatic takeoff for Paris on 20 May 1927 highlighted the need for variable-pitch propellers. (SI Nega- for cruising at altitude. With a fixed-pitch propeller, the setting of pitch involved tive No. A-4819A) an inherent compromise that somewhere sacrificed efficiency. The invention of the variable-pitch propeller bridged the gap between what could be considered the two major thrusts of the airplane design revolution of the interwar period: increased S. Hele-Shaw and T. E. Beacham patented a variable-pitch propeller mechanism in engine power and decreased drag. Great Britain in 1924, but it found little application, in part because the need for Nothing illustrated the need for a variable-pitch propeller more than Charles A. it was not as great as it was in America. The reason was topographical. Traversing Lindbergh’s dramatic takeoff for Paris on 20 May 1927. Clearing telephone wires at North America by air meant surmounting obstacles like the Appalachian and Rocky the end of Long Island’s Roosevelt Field by only 20 feet, Lindbergh’s ground-adjust- Mountains. In 1929, McCook (later Wright) Field engineer Frank Caldwell, who able propeller was set at a blade angle just below the optimum setting for cruise had designed most of the detachable-blade metal propellers in use, resigned from specifically to accommodate a better takeoff. Searching for every bit of economy the service to become chief engineer for Hamilton Standard Propeller Company. for the 3,610-mile transatlantic flight, Lindbergh almost did not get off the ground. (He moved into this position just as United Aircraft bought out Standard Steel and Document 3-13 features an April 1927 letter from Standard Steel Propeller Com- joined the two companies, forming Hamilton Standard.) As head of Hamilton’s pany to Ryan Airlines, a month before Lindbergh’s takeoff, which dealt with the production plant in Pittsburgh, Caldwell devoted himself to perfecting a control- critical pitch setting for Lindbergh’s historic Spirit of St. Louis Ryan-built airplane. lable-pitch propeller, which he accomplished by 1933. Called “the gear shift of the It was apparent to aeronautical engineers well before Lindbergh’s takeoff that air,” Caldwell’s device enabled pilots to adjust the pitch of their propeller blades a method to adjust the pitch of the propellers automatically in flight was needed automatically to match different flight conditions. This mechanism resulted in dra- to improve aircraft performance. But experience proved it was not easy to design a matic improvement in performance and won for Caldwell and Hamilton Standard mechanical system to accomplish this task reliably and as precisely as desired. Dr. H. the Collier Trophy in 1933 (see Document 3-14). Improvements in propellers con- 22 Chapter 3: The Design Revolution On a Streamline to the DC-3 23

The highly successful Lockheed Orion utilized retractable landing gear and an aerodynamically refined shape to generate unprecedented overall performance. Nevertheless, its wooden construction and lack of a high-lift flap system and a variable-pitch propeller kept it from being a truly revolutionary airplane. (SI Negative No. 1996- 0012; Videodisc No. 1B-14112)

before. The following year, 1924, the Verville-Sperry won the Pulitzer race, but at a speed of 215 mph, 30 mph slower than that achieved by Al Williams in his Curtiss Racer. Clearly, retractable landing gear represented a major aerodynamic advance, but it was not the only factor defining the performance of an aircraft. In terms of integrating a number of critical new design features, includ- The Verville-Sperry R-3 racer, winner of the 1924 Pulitzer Race, was an important precursor for “modern” high- ing retractable landing gear, the Boeing Monomail stands out. This airplane was performance aircraft with its streamline design and retractable landing gear. (NASM Videodisc No. 1B-95330) unprecedented in incorporating an all-metal structure, a smooth stressed skin, a cantilevered wing, and a Townend ring cowling. Unfortunately, it also possessed tinued, and by 1936, several new aircraft featured constant-speed propellers that an ineffective propeller that cancelled out most of the benefits of the lower drag. could automatically adjust pitch. The NACA contributed key research data during Because its speed range was so extensive, the Monomail needed a variable-pitch this period relevant to the design and operation of controllable-pitch propellers. propeller—a design then not yet available. Its fixed-pitch propeller could not match Another essential shelf item developed during the design revolution of the inter- the plane’s high speed and takeoff requirements. war period, one meant to eliminate a significant source of drag, was retractable land- In 1931, the Monomail was followed by a much more successful airplane ing gear. As Frenchman Louis Breguet had declared in his 1922 Royal Aeronautical employing retractable landing gear, the Lockheed Orion, a low-wing version of Society lecture on “Aerodynamical Efficiency and the Reduction of Transport Costs,” the pioneering Vega. (Also in 1931, the U.S. Navy procured its first aircraft with the undercarriage should be made to disappear (refer to Document 3-4), but it took retractable gear, the Grumman XFF-1 two-seat fighter.) In comparison to the fixed- a while longer before much happened to eliminate it. Some aviation historians have wing Lockheed Vega that came out just before, the advantages of the Orion’s retract- written that the first American aircraft to incorporate retractable landing gear was able gear were obvious: a maximum speed of 226 mph compared to 190 mph and the Boeing Monomail of 1930, but this is incorrect.12 In the 1923 Pulitzer Trophy a zero-lift drag coefficient of 0.0210 compared to 0.0278. These figures meant that Race at Mitchell Field, Long Island, the Army entered a low-wing monoplane, an the Orion could outperform any contemporary military aircraft. advanced version of the Verville-Sperry R-3, which incorporated retractable landing The use of retractable landing gear was delayed for so long primarily due to a gear. Mechanical problems forced the R-3 to drop out of the race, which was won general belief in the industry that such gear was too heavy for practical use. Com- by a sleek Curtiss Racer flown by Lt. Alford J. Williams at the then-terrific speed of pared to what appeared to be subtle aerodynamic benefits, the difficulties of design- 245.3 mph— almost 40 mph faster than the speed the race was won at only a year ing a reliable undercarriage that could be mechanically retracted into the body of an aircraft were glaringly obvious. More than any other single factor, it was the spec- tacular improvements realized with the Orion, which proved that the advantages far 12 See, for example, John D. Anderson, Jr., “Faster and Higher: The Quest for Speed and Power, “ in Milestones of Aviation (New York: Hugh Lauter Levin, 1989), pp. 105-106. The Dayton-Wright RB- superseded what actually amounted to small increases in weight, that put retract- 1 of the 1920s was also equipped with a landing gear that was semiretractable. able landing gear front and center of the shelf of new important new aircraft compo- 24 Chapter 3: The Design Revolution On a Streamline to the DC-3 25

landing gear. This was another development for which the NACA deserves considerable credit, and again, largely through the pioneering work of Fred Weick. As with many other aeronautical engineers in the 1920s and 1930s, Weick dreamed of building a low-cost and simple-to-fly airplane that was so inherently safe and inexpensive to operate and maintain that air travel in it could, for certain purposes, compete with automobiles as a mode of private and family transports. Weick pursued his dream in earnest in the early 1930s, when he and a small group of colleagues from NACA Langley designed the “W-1” in a private venture. This home-built experimental airplane had several unique features, including an elevator with upward travel that was limited to the point where the airplane could not be forced into a spin. Another innovative feature of the W-1 was its coordinated control system. Weick’s idea here was to reduce the number of controls from three to two by connecting the ailerons and rudder, thus eliminating the possibility of crossing these two controls and thereby simplifying the process of learning to fly. In order to make it easier both to taxi and land the plane, Weick and his associates also equipped the W-1 with what was then an unconventional undercarriage: Fred E. Weick’s W-1 home-built experimental airplane with steerable, tricycle gear precipitated a major change in they moved the two main wheels a short distance behind the plane’s center of grav- how engineers designed modern aircraft. (NASA Image No. L-11151) ity, took away the tail wheel, and added a nose wheel that could be steered. Such an arrangement, Weick thought, stood a very good chance of eliminating “ground nents. As new streamline designs started flying in the 1930s, the same fixed landing looping,” a serious problem even into the 1930s. All landplanes in production were gear that previously had accounted for only a small percentage of the overall drag still being equipped with tail skids—or, when paved runways came into use, with of an airplane, usually in biplane form, now accounted for a significant portion of a tail wheels. These gears had their main fixed-axis wheels located ahead of the center monoplane’s drag. As Laurence K. Loftin noted in Quest for Performance, “The con- of gravity, were naturally unstable directionally, and thus tended to ground loop. figuration and design details of the Lockheed Orion represented an extremely high Automobiles, bicycles, and motorcycles were stable in this regard because their main level of aerodynamic efficiency, a level that has seldom been exceeded in the years fixed-axis wheels were in back of the center of gravity. The conventional tail-wheel- since 1931.” Yet, as Loftin also pointed out, the Orion “lacked several features that type gears also nosed over easily, because the main wheels were just a little ahead later became an integral part of the propeller-driven aircraft in its final definitive of the center of gravity. Some of the early pusher airplanes, popular around 1910, form.”13 Like the Monomail, it also did not have a variable-pitch propeller, which had a single wheel well ahead of the center of gravity and two wheels in back of the meant its engine was not efficient over an entire range of flight conditions. This center of gravity, which took care of the nosing-over difficulty reasonably well. But limitation became clearer and clearer to aeronautical engineers in the early 1930s, all wheels were on fixed axes and could be steered very well on the ground. All of stimulating the development of the controllable-pitch propeller, without which the these realizations led Weick to conclude that the undercarriage of the W-1 needed full aerodynamic potential of a high performance, low-drag aircraft simply could to be quite different. Weick dubbed his new steerable undercarriage a “tricycle gear,” not be reached. The Orion also had only a simple trailing-edge flap, one designed to and it was a name that stuck. (In March 1932, Weick received a U.S. patent for his increase drag for approach and landing but not to increase maximum-lift coefficient. W-1 design with steerable tricycle gear; he received another patent for a two-control It really needed a high-lift system that increased maximum lift and reduced stalling airplane design incorporating a tricycle landing gear in March 1938; see Document speed. As suggested earlier, an advanced form of flaps was necessary—something 3-16). The gear performed extremely well for the W-1; the results were published that became standard equipment on high-performance aircraft by the late 1930s. in NACA reports. When first learning of Weick’s gear, some people in the industry Aerodynamically speaking, “retracting” technology was the key to landing gear said that it was just a reversion to the three-wheeled gears that had been used on (Document 3-15). But before leaving the subject of this particular “shelf compo- pusher airplanes before World War I, but this was not the case. Airplanes during nent,” it is important to note the critical importance of developing steerable, tricycle that era had fixed nose wheels, and operations took place in open grass fields, with a great deal of manpower available for handling the airplane. This meant that people in those days had not cared whether gear was stable in taxiing or not. There is no 13 loftin, Quest for Performance, pp. 89-90. 26 Chapter 3: The Design Revolution On a Streamline to the DC-3 27

of aeronautics at MIT, served as a consultant to this group, and he visited Langley Field several times to review the possibilities of applying the tricycle gear to such a large transport.14 The result of this activity was the one-and-only Douglas DC-4E, the first large transport to use the tricycle gear. Even then, the Douglas engineers made provisions for returning to a tail-wheel-type gear in case they were not entirely happy with the tricycle gear. During World War II, a modified version of the DC-4 was used in military activities as the C-54, complete with tricycle landing gear. The tricycle gear was also used on other military aircraft, including the Douglas A-20, Lockheed P-38, Bell P-39 and P-63, Consolidated B-24, North American B-25,

The one-and-only Douglas DC-4E represented many of the aerodynamic innovations that evolved during the 1920s and 1930s. It was the first large transport to use tricycle gear and its design process helped originate NACA research in the establishment of stability and control specifications prepared to insure good flying qualities. (SI Negative No. A0193510)

question that Weick’s tricycle gear was a new line of thought and a worthwhile improvement. At the NACA conference in 1935, Orville Wright had made exactly these points to Weick, which encouraged him greatly. NACA engineers found that with the later addition of wing fillets and a NACA cowling, the diminutive McDonnell Doodlebug experienced greatly reduced buffeting and aerodynamic interference. (SI Negative No. A43437-C) It was also in 1935 that engineers with the Douglas company showed the first interest in applying the tricycle gear to large airplanes. Dr. Arthur Raymond, chief Martin B-26, and Boeing B-29. It eventually became the standard gear for most all engineer at Douglas, sent F. R. Cohlbaum and W. Bailey Oswald to Langley to find military, airline, and general aviation airplanes. Even the NASA Space Shuttle, argu- out about the tricycle gear on the W-1 and to ask questions about what it might ably the world’s most sophisticated aircraft, benefits from the same type of tricycle do for larger airplanes. After their visit, Douglas, in cooperation with the Army, gear with steerable and castered nose wheel that Weick designed back in the 1930s.15 tried the tricycle gear on one of its Dolphin airplanes, which originally possessed a Again, this development did not relate directly to improved aerodynamics, but it tail-wheel-type gear. The Dolphin was an amphibian flying boat, and the Douglas shows how various new design features synergized in the airplane design revolution engineers simply moved the wheels of the main gear back a bit and put a castering of the interwar years. nose wheel under the front of the hull. These tests confirmed the gear’s advantages. Another part of this synergy came in the form of wing fillets, which were con- In October 1935 TransContinental and Western Air (as TWA was known before cave fairings used to smooth the interior angle of the wing-fuselage juncture. With- 1950) also asked for information about the possible use of tricycle gear on a transport airplane. The NACA responded by pointing out the various advantages of the gear, plus the fact that with a twin-engine transport, one could extend the forward 14 For an account of Edward P. Warner’s outstanding career in American aviation, see Roger E. Bilstein, part of the fuselage sufficiently to support the nose gear well forward and also pro- “Edward Pearson Warner and the New Air Age,” in Aviation’s Golden Age: Portraits from the 1920s vide a satisfactorily long wheelbase. and 1930s, ed. William M. Leary (Iowa City IA: University of Iowa Press, 1989), pp. 113-126. In Soon thereafter, three or four airlines met with Douglas in the hope of getting a various ways, technically and politically, Warner made notable contributions to the “reinvention” of the airplane. transport that was larger than even the new DC-3 that had just appeared. Edward P. 15 For Weick’s account of his development of the tricycle gear for the W-1 airplane, see Weick and Warner, a former NACA chief physicist, editor of Aviation magazine, and professor Hansen, From the Ground Up, pp. 131-140. 28 Chapter 3: The Design Revolution On a Streamline to the DC-3 29

out fillets, the intersection of wing and fuselage experienced airflow interference. A smooth “fairing” of the area where the wing joined the fuselage (i.e., fitting and shaping so as to make it smooth and streamlined) greatly improved the handling characteristics and efficiency of low-wing monoplanes. Developed at the Guggen- heim Aeronautical Laboratory of the California Institute of Technology under the direction of Theodore von Kármán, the fillet soon became a standard feature of monoplane designs (see Document 3-17). Both the Boeing Monomail of 1930 and the Northrop Alpha of 1931 benefited from wing fillets, as would the culminating airplane of the design revolution, the Douglas DC-3. As for the Alpha, the advantages of wing fillets could not overcome its basic aerodynamic obstacles, notably a fixed landing gear and an open cockpit for the pilot. It is curious that so many new aircraft during this era combined anachronistic features with important new innovations. But such combinations of old and new are, in fact, typical of transitory times before all the features of a new technology come together and form a new design paradigm. Equally typical of the transitory period is the appearance of a “mainstay design,” one that maximized the existing state of the art but offered a sophisticated overall product that nonetheless augured important elements about the future. Such an aircraft was the ubiquitous Ford Trimotor, a 13- to 15-passenger transport that did the lion’s share of work for America’s budding airlines in the late 1920s and early 1930s. Although the Trimotor remained aerodynamically primitive in some respects, such Aircraft such as the Lockheed Sirius “Tingmissartoq” flown by Charles and Anne Lindbergh represented the as its fixed landing gear and uncowled engine nacelles, aerodynamic refinement, as transitional nature of aerodynamic design with its advanced streamline shape combined with an open cockpit, ground-adjustable pitch propeller, and large floats. (SI Negative No. A-5172) Document 3-18 shows, had been one of the goals of Ford Motor Company’s Air- craft Engineering Department when it started designing the Trimotor from 1925 to 1926. Henry Ford had built a mammoth reputation as a manufacturing and production genius in the automobile industry, and he wanted to do something similar

The Ford Trimotor 5-AT-B was a “mainstay design” for America’s growing commercial airline industry, but its improperly placed engine nacelles below the wing, angular windscreen, exposed engines, ground-adjustable pitch propellers, fixed landing gear, and corrugated aluminum construction did not reflect the rapid advances in The Martin B-10 represented the revolution in combat airplane design, and in key respects, presaged the total aerodynamic streamline design. (SI Negative No. 47937-J) package of a “reinvented” airplane. (SI Negative No. 78-1328) 30 Chapter 3: The Design Revolution On a Streamline to the DC-3 31

design and in key respects presaged the total package of a “reinvented” airplane— the Martin B-10 bomber. But the aircraft that integrated all the most desirable features emerging from the airplane design revolution even more fully was the Boeing 247. Aviation historians consider this Boeing design, first flown in February 1933, to be the first “modern airliner.” The definitive version of the airplane, the model 247D, had the following features: all-metal, stressed-skin construction; cantilever wings; retractable landing gear; an efficiently cowled, lightweight radial engine; controllable-pitch propellers; and single-speed geared supercharger. Model 247D was also the first transport aircraft to employ rubber de-icing boots and to have a significant amount of instrumentation for blind flying. Aerodynamically, the design was quite far advanced, with split-type landing flaps and a fairly high wing loading. Its 0.0212 CD,O was rather low, and its 13.5 maximum lift-drag ratio was quite high; both numbers compared favorably with the performance of virtually all previous aircraft. Operating for the airlines, the Boeing 247 carried The Boeing 247D integrated all the most desirable features emerging from the airplane design revolution, making it the first “modern airliner.” (SI Negative No. 75-12118) up to ten passengers comfortably, at a cruising speed of 185 mph at 7,500 with airplanes. So resolved, he put the resources of his entire company behind the feet. In service mainly with United building of a Ford transport. He even bought out the Stout Engineering Company, Airlines(along with Boeing, its maker, which had recent experience with producing an innovative, cantilever-wing trans- part of the United Aircraft and Trans- port design. Stout’s transport, which was also all metal, would soon become famous port Corporation), the airplane estab- for running on the Stout-Ford airlines between Chicago, Detroit, and Cleveland. lished a new standard for commercial Ford would soon have William B. Stout himself at work, in his small engineering air travel. In doing so, it provoked other aircraft manufacturers, as well as Jack Frye’s August 1932 letter to American aircraft laboratory in Dearborn, Michigan, on the design of the Ford Trimotor. manufacturers requesting a new all-metal airliner for Two other airplanes that represented a link between old and new were the previ- the airline operators, to seek the design TWA led to the creation of the Douglas DC-series air- ously mentioned Lockheed Vega, which appeared within one year of the Ford Tri- of their own more refined aircraft. craft. (SI Negative No. 75-5208) motor, and the Lockheed Sirius of 1930. All three planes reflect the flux and inde- One aviation executive who des- terminacy of aerodynamic development of aircraft during the late 1920s, before all perately sought a new airplane was Jack Frye, vice president for operations of Trans- of the components of the airplane design revolution came together into a coherent continental and Western Airlines (TWA), a company that was precluded from whole. Synergy of all the necessary shelf items was not yet possible, because not all buying the new Boeing 247 because of the corporate connection between Boeing the shelf items had yet appeared, so to speak, “on the shelf.” But all three designs and United Airlines. On 2 August 1932, Frye sent letters to Curtiss-Wright, Ford, indicate that their designers were, without question, paying attention to reducing Martin, Consolidated, and Douglas seeking proposals for a new all-metal trimo- aerodynamic drag through streamlining. Harold Hicks replaced George Pruden as tor design, one that he hoped could profitably replace TWA’s aging fleet of Fokker the head of Ford’s Aircraft Engineering Department (Henry Ford fired Pruden for a trimotors. That letter, reproduced in its entirety as Document 3-21, reflected Frye’s breach of company etiquette; he appeared in a newspaper photo as a representative belief that trimotored aircraft were still the norm for airline operation. Interestingly, of the Ford Company) and oversaw the development of the trimotors. Document the TWA leader stipulated no other design details and only suggested a few of the 3-19 concerns the design of the Sirius airplane. Document 3-20 details yet another desired performance characteristics. He left it up to the manufacturers to establish one of these transition types—one that stimulated a revolution in combat airplane the formal design parameters. 32 Chapter 3: The Design Revolution On a Streamline to the DC-3 33

Intrigued by the challenge of producing a design that might even turn out KLM finished second in the 11,300-mile London-to-Melbourne MacRobertson better than the Boeing 247, Donald W. Douglas and his design team in Santa Trophy Race—and amazingly did so with a full load of mail and three passengers. Monica, California, began work on a 12-passenger twin-engine airliner incorporat- Quickly, KLM ordered a dozen DC-2s, followed by numerous orders from other ing all the latest advances in aircraft components and design. This included certain European carriers. For the first time, an American airliner began to outcompete the features that the Boeing 247 lacked, notably variable-pitch propellers and split flaps Fokker transports, which had dominated European airline service for several years. for higher wing loading. A Douglas delegation headed by a young Caltech gradu- Although a British plane had won the MacRobertson Race, the winner had been a ate, Arthur E. Raymond, traveled to TWA’s offices in New York City and won the special customized racer, the De Havilland D.H.88 Comet. British aviation lead- contract for the new airplane. After ten months of development work, the first ers recognized that the real significance of the race lay in the American commercial of the new Douglas DC series transports, the DC-1, took to the air in July 1933. airliner coming in second. The editor of London’s Saturday Review declared soon The prototype all-metal DC-1 incorporated the most advanced innovations of the after the race: “Britain has won the greatest air race in history, but she has yet to start time. In addition to the newest shelf items, the Douglas design team utilized the on an even greater air race: a race in commercial and military supremacy.” Sadly, no latest in aeronautical design methods to fully integrate them into the overall design. British airplane, not even the best machine in regular use with the Royal Air Force This utilization included an unprecedented level of “research for design” conducted “at the present time is fast enough to have finished the race within a thousand miles within a wind tunnel (see Document 3-22). of the American machines.” (A Boeing 247 flown by Roscoe Turner and Clyde It was not just technological incentive that led to the DC-1’s innovative design. Pangborn finished third in the race, two and a half hours behind the DC-2.) “It is Economic incentives further hastened technical innovation when Douglas engineers almost incredible, but it is true,” the British newspaperman lamented.16 elongated the fuselage of the DC-1 prototype (see Document 3-23) to allow for two In the design of the Douglas DC-2, the two main approaches that American more passengers, bringing the capacity up to 14. The resultant DC-2 first flew on engineering had been following for the purpose of streamlining aircraft designs (i.e., 11 May 1934 and became the first commercially available Douglas transport. Flying commercial efficiency and greater speed) converged very successfully, and American TWA’s 18-hour “Sky Chief” transcontinental route from Newark to Los Angeles aviation benefited greatly from it, on a worldwide scale. The DC-2 clearly outper- (with en route stops at Chicago, Kansas City, and Albuquerque) at speeds up to formed all other airliners. For the first time, American industry was producing a 175 mph, the DC-2 carried the same amount of passengers as the old trimotors transport so superior to all others available that even European carriers commit- at less cost and at faster speeds. The success of the DC-2 was not just national but ted to their own national aviation industries had no choice but to start flying the international. It was also stunning. In 1934, a DC-2 flown by the Dutch airline American planes. The significance of the DC-2 does not end there. As historian Richard K. Smith noted, the appearance of the DC-2 also created, for the first time, “a distinct division of labor between the design of military and civil aircraft.” Before 1933, bombers were typically just converted airliners with bomb shackles, armor, and guns added. But the new streamlined forms dictated more specialized designs. Streamlining called for bombs to be carried internally, and because bombs took less space than a load of passengers, from this point on “cross-section designs of bombers and airline equipment would move in opposite directions.”17 In this sense, the reinvention of the airplane brought on greater specialization. Streamlining was a general objective, but the special forms of streamlining required by aircraft with significantly different missions led to one of the clearest bifurcations in aircraft design history. This was a critical element of change that is often missed when discussing the airplane design revolution of the interwar years. As mentioned earlier, the DC-2 design incorporated the latest refinements enjoyed by the Boeing 247, plus a higher wing loading and split-type flaps. The Douglas Company found itself in the enviable position of trying to fill an over-

16 Dutch KLM’s use of the DC-2 in the 1934 London-to-Melbourne MacRobertson Trophy Race announced to the Quoted in Terry Gwynn-Jones, “Farther: The Quest for Distance,” inMilestones of Aviation, p. 67. world the growing ascendancy of American aeronautical technology. (SI Negative No. 523183) 17 Richard K. Smith, “Better: The Quest for Excellence,” in Milestones of Aviation, p. 253. 34 Chapter 3: The Design Revolution On a Streamline to the DC-3 35

our Destination Document, it was the first airplane that enabled the airlines to make money by solely carrying passengers. Technologically, it represented a completed synergy of virtually all the major innovations that had taken place in aircraft design since 1920. Aerodynami- cally, the airplane was extremely far advanced, producing a very low CD,O of 0.0249; this was 17 percent less efficient than the Boeing 247, which had an extremely low value of 0.0212. But this difference was because the DC-3 had a much larger fuselage—to accommodate three-abreast seating—that resulted in a larger ratio of wetted area (for wings and wing-like surfaces, the wetted area is related C.R. Smith’s need for new aircraft for Ameri- to the exposed planform area) to wing area. can Airlines resulted in the DC-3, which Smith declared as the first airliner to enable the air- The highly successful DC series of aircraft, such as the Douglas Skysleeper Transport designed for American The DC-3 enjoyed a very high L/D, a ratio airlines, garnered Douglas Aircraft the 1935 Collier Trophy. (NASM File No. AD-761192-75) lines to make money by solely carrying pas- of 14.7. This L/D was higher than anything sengers. (SI Negative No. 8094; Videodisc that came before—and better than virtually No. 2B-75566) whelming backlog of orders for the plane. Designing an even bigger and more effec- all that came after, certainly with propellers. tive transport became a growing possibility when Cyrus Rowlett “C.R.” Smith of (Only the Boeing B-29 bomber of 1944 and the Lockheed L.1049G Super Con- American Airlines, the third U.S. transcontinental service (with TWA and United) stellation airliner of the early 1950s enjoyed higher L/Ds, of 16.8 and 16.0, respec- wanted an aircraft to replace the fleet of aging Curtiss Condor biplanes that Ameri- tively.) One very noteworthy aerodynamic feature of the DC-3 was its sweptback can had been using on its nighttime Pullman-style sleeper service. American’s chief wing. This was not engineered for the same reasons that wings of transonic and engineer, William Littlewood, determined that by widening the fuselage of the DC- supersonic aircraft would later be given sweepback; rather, it had to do with posi- 2 by 26 inches and adding 10 feet to its wingspan, it could accommodate 14 sleep- tioning the airplane’s aerodynamic center properly in respect to its center of gravity. ing berths. The revised airplane, known as the Douglas Skysleeper Transport, made To some extent, this had been done also with the DC-2. But as the design of the its maiden flight on 17 December 1935, in celebration of the 32nd anniversary of DC-3 evolved, it became evident to Douglas’s engineers that the center of gravity the Wrights’ flight at Kitty Hawk. Fitted with Wright R-1820 engines and outfitted was located farther toward the rear (or aft) than they expected. They mounted the for 21 daytime passengers (without the sleepers), the airplane flew as the DC-3. On outer panels of the DC-3 wing with greater sweepback, moving the aerodynamic 1 July 1936, President Franklin D. Roosevelt, on behalf of the National Aeronautic center to the correct position. This change helped the airplane achieve a cruising Association, presented the Robert J. Collier Trophy to the Douglas Company for speed of 185 mph (at 10,000 feet) when carrying a full load of 21 passengers. In the greatest achievement in American aviation in 1935, the refinement of the DC other words, the DC-3 benefited from an even more streamlined shape than its series aircraft. In his remarks, President Roosevelt said, “This airplane by reason of predecessors, while increasing capacity by 50 percent. Thanks not only to streamlin- its high speed, economy, and quiet passenger comfort has been generally adopted by ing but also to the entire process of reinventing the airplane, for the first time, the transport lines throughout the United States. Its merit has been further recognized American airline industry had a passenger aircraft that could make a profit without 18 by its adoption abroad, and its influence on foreign design is already apparent.” a government airmail subsidy. Soon, the DC-3 became the most popular and reliable propeller-driven airliner Without a doubt, 1934, 1935, and 1936 were banner years for both American in aviation history. Its appearance marked the culmination of the design revolution and global aviation. From the point of view of aviation only, it has been argued that in American aircraft and, as C. R. Smith, president of American Airlines, noted in it was the aircraft developments associated with this period that, as much as anything else technological contributed directly by American industry, helped begin to lift the U.S. economy out of the Great Depression; aircraft developments certainly 18 Quoted in Douglas J. Ingells, The Plane that Changed the World: A Biography of the DC-3 (Fallbrook, CA: Aero Publishers, Inc.), p. 166. played a part in the general recovery that ultimately depended on involvement in 36 Chapter 3: The Design Revolution On a Streamline to the DC-3 37 the Second World War. By the mid-1930s, the streamlined shape defined largely by The NACA had conducted many studies related to stall before the mid-1930s, the American aeronautics community became standard in all new aircraft designs, but the operating problems of a large new airliner as revolutionary as the DC-3 making advance performance through design refinement a reality and the U.S. air- sparked a new wave of concern. In September 1937, United Airlines loaned NACA craft industry the world leader (see Document 3-24). As seen in this chapter and its Langley one of its DC-3 Mainliners in order to conduct stall tests (see Document documents, a myriad of new flight technologies contributed to this airplane design 3-25). Out of this specific program came a more general investigation into stalling. revolution. Although major developments in propulsion, structures and materials, Through the 1940s, the NACA worked on various stall-warning indicators and flight instrumentation, and stability and control technology played fundamental continued to look generally into stall phenomena. In the early 1940s, as readers will roles in the reinvention of the airplane, the role of aerodynamic refinement was learn in a later document in this chapter, the NACA also played the key role in spell- primary in many ways. The introduction of superior propellers with controllable ing out “stalling characteristics” as part of a larger program of establishing uniform pitch, cantilever wings, wing fillets, flaps, efficient engine cowlings, proper nacelle “flying qualities” requirements for American aircraft. placement, and retractable landing gear all pushed the airplane to extraordinary new In flying United Airlines’ DC-3 Mainliner in late 1937, NACA Lang- levels of sophistication and performance. So, too, did the definition of advanced ley investigated more than the stalling problem; it also looked into the equally airfoils, perhaps the most important single component contributing to the airplane if not more dangerous problem of aircraft icing. As historian Glenn Bugos has design revolution. This was such a crucial topic for aerodynamicists, that it will be explained, icing was a critical systems-wide problem for aircraft, then and now: explored separately as the entire focus of the next chapter. It took more than just the integration of various airplane components to cement Ice caused aircraft to crash by adding weight and preventing the pilot the airplane design revolution within professional aeronautical practice and pre- from climbing above the icing clouds, so that the aircraft gradually lost pare aeronautical engineers for the future. Even after the reinvented airplane had altitude and slammed into the ground . . . [I]ce accreted along the wing been made essentially complete in the form of the DC-3, significant problems still and tail leading edges disturbing lift and adding drag. Ice clogged the inter- affected total airplane performance. As Document 3-25 shows, one of these prob- stices of rudders and ailerons, preventing control and inducing buffeting. lems was aerodynamic stall, the potentially dangerous, even fatal, flight condition It changed the aerodynamic profile of the propeller, causing it to vibrate when an aircraft started to fly at an angle of attack greater than the angle of maxi- and exert less thrust per horsepower. It coated windshields, so the pilot flew mum lift, resulting in a loss of lift and an increase in drag. Stall had always been a blind. Ice made antenna wires oscillate and snap, and generated static that concern for engineers and pilots, but in the early years of aviation, stalling speeds rendered useless most radio communication and navigation. It distorted had remained low, in the range of 40 to 50 mph. This was due to the relatively poor pitot shapes, so that pilots got erroneous airspeed readings. And it clogged flying characteristics of aircraft of that period, the fact they did not benefit from carburetors, suffocating the engine. high-lift devices, and related directly to the short, unpaved fields then serving as airports. A pilot simply could not come in for a landing at a very high speed and Within minutes, pilots could lose all of their critical systems, not just the engine, expect to land safely. wings, control surfaces, indicators, and radio, but also their own sight as well.19 The design revolution of the interwar years changed the descent and landing Icing was a problem to which the NACA committed considerable time and situation dramatically. The flying characteristics of airplanes improved greatly. High- energy. A rash of aircraft accidents traced to icing problems attracted public atten- lift devices became sophisticated and commonplace; runways became paved and tion in the late 1930s; a number of commercial operators, not just United Airlines, quite long. Aircraft could manage landings at much higher speeds; but in doing clamored for useful information on the subject. The NACA launched a compre- so, they ran new risks of stalling not experienced earlier. The stalling speed of the hensive study of the icing problem that would last for many years. Researchers DC-3 rose to over 60 mph; the Boeing B-17 to over 80; the North American P-51 looked for mechanical, chemical, and thermal ways of breaking up ice before it to about 100; the Martin B-26F’s stalling speed rose to over 120 mph. By the start formed dangerously on an airplane’s vital surfaces. Innovative new tests looking of World War II, the stalling speeds of virtually all advanced aircraft had risen into for answers to icing began in flight and in wind tunnels. Various de-icing systems the range of 80 to 100 mph; without the help of high-lift devices, stalling speeds were tried, including a thermal (heat-directed) system for which the NACA and its would, in fact, have become much higher, further increasing risk. Increased speeds generally equated to progress in aviation, but landing at higher and higher speeds, even with the aid of high-lift devices, raised new issues that aeronautical engineers 19 glenn Bugos, “Lewis Rodert, Epistemological Liaison and Thermal De-Icing at Ames,” in From had to resolve. Engineering Science to Big Science: The NACA and NASA Collier Trophy Research Project Winners, ed. Pamela E. Mack (Washington, DC: NASA SP-4219, 1998), p. 32). 38 Chapter 3: The Design Revolution On a Streamline to the DC-3 39

engineer Lewis Rodert won the Collier Trophy for 1946. Document 3-26 surveys and stresses on combat aircraft while dive-bombing. In the 1930s, NACA research- the NACA’s pioneering work on aircraft icing from the late 1930s to 1948. It was a ers worked out charts showing the relationships between dive angle, speed, and research specialization that NASA would pursue with vigor. the angle required for recovery. Using these charts, the Navy established design Another generic problem that the NACA investigated (to the advantage of the requirements for its dive-bombers around 1935. The NACA later conducted other airplane design revolution of the 1930s) centered on gust loads and gust allevia- flight dive tests meant to ensure that elevator control force would not hinder the tion. One event sparking early interest in this problem was the destruction of the acceleration a pilot needed to recover his aircraft from a dive. Through this research ZR-1 Navy airship Shenandoah, which crashed in a thunderstorm near Ava, Ohio, program, the U.S. aeronautics community came to a much better understanding of on 3 September 1925, killing 14 of 43 on board. Engineers in the PRT section at the distribution of loads between the various parts of the airplane—wing, fuselage, Langley reacted to this tragedy by mounting some rudimentary equipment on the and tail surfaces—and how these individual parts were “loaded” by the wide range top of their tunnel building trying to test the magnitude of gusts found in the free of maneuvers, some of them extreme undertakings by a high-performance aircraft. air. Later, NACA researchers collected pressure distribution measurements of air As already suggested in reference to defining “stalling characteristics,” another loads in wind tunnels and in flight, but there was still no strong impetus to apply major NACA contribution of this period was its specifying of uniform “flying qual- them to the design of specific aircraft. Not until the early 1930s did a coordinated ity” requirements for American aircraft (see Document 3-28). As already empha- NACA program of gust research begin that was interested in applying such data sized, the airplane design revolution of the interwar period involved more than to aircraft design. A few special NACA committees then formed to oversee loads the development of aircraft component technology and its various integrations and research. Eventually, Langley created a separate structures research division and, in synergies. It also involved broader intellectual developments—notably the transla- 1938, built a structures research laboratory. tion of ill-defined problems, issues understood and handled previously in a largely As this commitment materialized, it grew clearer and clearer to everyone that the qualitative and subjective manner, into much more objective problems defined and stresses brought on by gusts had to be a more essential factor in the design of aircraft, resolved on a stronger quantitative basis. In this respect, the reinvention of the air- especially civil airliners. Understanding gust loads lay somewhere about midway plane hinged on an epistemological reorientation within American engineering, one between aerodynamics and meteorology, and without question threatened the safe that changed not only what aeronautical engineers knew but how they knew it. and comfortable operation of aircraft, with special concern for passenger transports. In retirement, as part of a productive second career as a historian of technology, As an aerodynamic problem, the NACA first conducted gust loads research “on the former NACA/NASA research engineer Walter G. Vincenti delved deeply into this premise that a gust acted almost like an instantaneous change of attack of the air- epistemological transition. Several chapters in his prizewinning 1990 book, What plane encountering it.”20 Engineers who came to specialize in the problem, notably Engineers Know and How They Know It: Analytical Studies from Aeronautical History, Langley’s Richard V. Rhode (a 1925 University of Wisconsin graduate who would looked into the significance of this change, how and why it happened, as well as win the Wright Brothers Medal of 1935 for his gust loads research) later refined its impact on the wider world of aviation. In particular, his chapter, “Establish- this concept by taking into consideration the distance over which a gust extended ment of Design Requirements: Flying Quality Specifications for American Aircraft, and the time it took for the wing flow to adjust to the new angle of attack. Rhode’s 1915-1943,” analyzed the complex intellectual and social process by which U.S. concept of a “sharp edge” gust became “the backbone for all gust research.”21 Docu- aeronautical engineers in the 1920s and 1930s came to establish more objective ment 3-27 provides a string of documents from the 1930s concerning the emerging specifications for aircraft design. importance of this major new field of research. Aeronautical engineers as late as the start of World War II really did not know how to phrase, let alone categorize, their observations about the performance of Although not covered by any documents in this chapter, the NACA’s study an airplane. Everyone associated with aircraft development and operation wanted of gust loads for the purpose of articulating design requirements carried over into good flying and handling characteristics. They wanted machines that precisely, rap- several other critical areas, some of which concerned military more than civilian idly, and predictably obeyed a pilot’s inputs—without unwanted dips or drops and aircraft. One of these areas involved a systematic exploration, both theoretical and without the pilot having to manhandle controls. Pilots, however, had a hard time experimental, into the dangers of aerodynamic flutter. Another concerned the loads putting what they wanted into words. For example, a pilot reported his machine as “tail heavy” or commented on the “lightness” of his controls, but nowhere were any of these terms defined concretely. Pilots described what “flying qualities” were most 20 Hartley A. Soulé, “Synopsis of the History of the Langley Research Center, 1915-1939,” HQA acceptable to them impressionistically, in their own words, to their own apparent HHN-40, 1966, p. 39. satisfaction, but without a clear, quantitative basis. Whatever was actually tangible 21 Soulé, “Synopsis,” p. 39. 40 Chapter 3: The Design Revolution On a Streamline to the DC-3 41

in the relationship between the “feel” of an aircraft and its actual nuts-and-bolts (5) rudder control characteristics; performance remained vague and imprecise. (6) yawing moment due to sideslip; Vincenti’s chapter revealed the 25-year process it took, from 1918 to 1943, (7) crosswind force characteristics; before “flying qualities” became well enough defined to be specified. As with so (8) pitching moments due to sideslip; and many elements of the airplane design revolution, what unfolds in Vincenti’s story (9) uncontrolled lateral and directional motion. is the history of an idea: “The notion that specifications could usefully be written for something as subjectively perceived as flying qualities had itself to be realized The third category, “stalling characteristics,” has been discussed earlier in this intellectually and verified in the real world. It was not at all an obvious or obviously chapter in relation to the Douglas DC-3. In his book chapter on the “Establish- useful idea at the outset.”22 By the early 1940s, however, specifications did become ment of Design Requirements,” Vincenti looked at the NACA’s articulation of all defined, as did the whole concept of “flying qualities.” NACA test pilot Melvin the sub-divisions, with a specific focus on elevator control in steady flight, uncon- N. Gough defined that concept early in World War II as “the stability and control trolled longitudinal motion, and elevator control in accelerated flight, all part of the characteristics that have an important bearing on the safety of flight and on the “requirements for longitudinal stability and control.”24 pilot’s impressions of the ease and precision with which the aircraft may be flown As Document 3-28 indicates, the immediate catalyst for the NACA’s flying- and maneuvered.”23 quality research came in 1935 and resulted from a set of specifications prepared for The string of documents in Document 3-28 all relate to the research program the Douglas DC-4E. Edward P. Warner, the accomplished aeronautical researcher undertaken by the NACA from the mid-1930s onward to establish precise “flying and former editor of Aviation who was then working as a consultant for the Douglas qualities.” In essence what the NACA did, primarily through flight research of Aircraft Company, asked the NACA to help him specify the stability and control full-scale aircraft, was measure how aircraft responded to specific control inputs characteristics to be built into the DC-4 transport. In December 1935, the NACA and then correlate them with pilots’ opinions of the aircraft’s behavior. The NACA Aerodynamics Committee, which was chaired by Warner, approved what became investigators then related what they found to the engineering parameters that had Research Authorization (RA) 509, “Preliminary Study of Control Requirements been employed in designing the aircraft. The result was a long menu of “require- for Large Transport Airplanes.” The purpose of this investigation was to determine ments” essentially falling into three main categories. The first group, “requirements “what specific qualities pilots desired, so that they could be numerically specified in for longitudinal stability and control,” subdivided into: future design competitions.” A team of NACA Langley researchers under Hartley A. (1) elevator control and takeoff; Soulé started this work in 1936 using a Stinson cabin monoplane. Langley instru- (2) elevator control in steady flight; mented the airplane so that its response characteristics, following known control (3) longitudinal trimming device; inputs from the test pilot, could be measured, related to design parameters, and cor- (4) elevator control in accelerated flight; related with the pilot’s qualitative evaluation of the ease and precision with which (5) uncontrolled longitudinal motion; he maneuvered the plane. Soulé’s team continued this effort using all 12 different (6) limits of trim due to power and flaps; and airplanes that could be obtained for this purpose until 1941, when it was ready to (7) elevator control and spinning. specify its first set of uniform requirements for the flying qualities of airplanes. The U.S. Army and Navy enthusiastically welcomed the NACA’s product and The second category, “requirements for lateral stability and control,” subdivided into: started to revise the preliminary specifications to meet their immediate require- (1) aileron control characteristics; ments. In 1942, both services asked the NACA to continue validating and upgrad- (2) yaw due to ailerons; ing handling requirements specifically for military aircraft. This request made many (3) rudder and aileron trim devices; more airplanes available to the NACA as well as a broader view than would have (4) limits of rolling moment due to sideslip; been possible with just a few select commercial aircraft. By the end of World War II, the NACA had measured the stability and flying qualities of 60 different aircraft, and military and civil aircraft handling requirements had been standardized. This 22 Walter G. Vincenti, “Establishment of Design Requirements: Flying Quality Specifications for effort foreshadowed the extensive work that would be undertaken in the field over American Aircraft, 1918-1943,” in What Engineers Know and How They Know It: Analytical Studies the next five decades, leading to the present uniform rating system. Unquestionably, from Aeronautical History (Baltimore and London: The Johns Hopkins University Press, 1990), p. 52. 23 Quoted in Loftin, Quest for Performance, p. 117. 24 Vincenti, “Establishment of Design Requirements,” pp. 93-98. 42 Chapter 3: The Design Revolution On a Streamline to the DC-3 43

the appearance of more objectively defined flying qualities proved to be a culminat- falo. The Bureau of Aeronautics wanted the NACA to look for “ kinks” or “bugs” ing development in the airplane design revolution, one that led to a more concrete in the plane’s general design and to determine, in only one week’s time, what drag understanding of aircraft operation, a sounder approach to aircraft design, and a reduction could be expected from changes that might readily be incorporated in much improved sense of the pilot-machine interface. the event that the XF2A was put into production. The NACA quickly agreed, and Not all revolutionary episodes turn out so neat and tidy, but one can say that even before a formal research authorization was transmitted to Langley laboratory, the airplane design revolution ended with a literal “dusting off” of what had become the NACA flew an XF2A to Langley Field for tests in the FST. The FST team at of the modern airplane, at least in its sleek military forms. Within the context of Langley acted quickly to satisfy the Navy’s urgent request, and within a few weeks NACA research, this final stage took the form of “drag cleanup,” a program of sys- had discovered ways to increase the Buffalo’s speed by 31 mph to 281, more than a tematic aerodynamic refinement involving dozens of aircraft that were to see action 10 percent improvement in performance. in World War II. Word got around, and soon not just the Navy but the Army as well was send- One might think that with the appearance of the revolutionary DC-3 that the ing all of its new prototypes to Langley for drag cleanup. Between April 1938 and aerodynamics of propeller-driven aircraft had become so refined that no further November 1940, Langley gave 18 different military prototypes thorough goings- improvements could be made. An ultrasleek internally braced monoplane with over in the FST to see if the airplanes could be improved in any way. On this list was retractable landing gear like the DC-3 could be expected to enjoy something akin the Grumman XF4F-2 Wildcat (June 1938), Vought-Sikorsky SB2U-1 Vindicator to “ideal” aerodynamics, but in actual service, that hardly proved to be the case. (August 1938), Curtiss P-36A Mohawk (August 1938), Curtiss XP-40 Kittyhawk Ideally, in aerodynamic terms, the drag of such an airplane—more specifically, its (August 1938), Grumman XF4F-3 Wildcat (September 1939), Republic XP-47 CD,O, defined earlier—should be extremely low, “only slightly in excess of that Thunderbolt (May 1940), Chance Vought XF4U-1 Corsair (September 1940), and which would be calculated with the use of the total wetted area of the airplane and Consolidated XB-32 Dominator (November 1940). Through its systematic drag a skin friction corresponding to a turbulent boundary layer.”25 But, as facts murder reduction investigations in Langley’s FST, the NACA did its best to help industry theory, real-world circumstances tended to tarnish and undermine this ideal. In realize some dramatic increases of speed in its production aircraft. This effort can be actual service, not even aircraft like the DC-3 ever achieved their ideal drag coef- seen clearly in the contents of both Document 3-29, which concerns drag cleanup ficient. Roughness or unevenness on an aircraft’s surface, even smashed insects on of the XF2A, XP-40, and XP-41, and Document 3-30, involving drag cleanup of a wing, disturbed smooth airflow and increased drag. So, too, did antenna or guns the Bell XP-39 Airacobra. Although the history of the XP-39 Army fighter plane’s projecting outside the smooth basic contour of an airplane. Even air unintention- top speed was complicated by a series of changes related to its engine, armor, and ally leaking through an aircraft structure influenced aerodynamic performance in guns (which significantly increased the machine’s overall weight, and thus made adverse ways. Sometimes, an aircraft in actual service experienced twice as much speed increases difficult), careful analysis confirms that if the XP-39 had not gone drag as its calculated ideal (based on wetted area and a turbulent skin friction factor). through drag testing, the production Airacobra’s top speed of just under 370 mph This was true, for example, of the Messerschmitt 109 fighter, the archenemy of the would have been quite a bit lower, in the range of 25 to 30 mph.26 Spitfire and the Hurricane during the Battle of Britain in 1940. A great deal more drag cleanup work was started when a second NACA labora- What this underscored was the absolute need for “detailed design” work if actual tory—the Ames Aeronautical Laboratory—opened for business at Moffett Field drag were to come anywhere close to ideal values. Everything about an airplane in Sunnyvale, California, near San Francisco, from 1940 to 1941. With so much and its parts required meticulous attention. In the early 1930s, experience in the of the aircraft industry located on the West Coast, it was quicker and easier to test NACA’s 30-by-60-foot Full-Scale Tunnel (FST) at Langley indicated that even the the new planes at Ames rather than at Langley. Thus, in addition to a new family most minute external protuberances could seriously undermine aerodynamic per- of high-speed wind tunnels, at Ames, the NACA constructed a 7- by 10-foot wind formance, but it took a desire to obtain every last ounce of benefit from the airplane tunnel and a 40- by 80-foot wind tunnel that also began to serve as workhorse facili- design revolution, plus the threat of war, before the U.S. aeronautics community, ties in the comprehensive drag cleanup program. specifically the U.S. military, sought systematic “drag cleanup” for their aircraft. In terms of the history of aerodynamics, compared to bold new inventions or What first prompted the NACA’s systematic drag cleanup program was a loud ingenious new theories, the significance of the NACA’s drag cleanup work might cry for help from the U.S. Navy, which in 1938 was unhappy with the 250-mph seem pedestrian. But this sort of systematic experimental investigation should not flight test performance of its new experimental fighter, the Brewster XF2A Buf-

26 See James R. Hansen, Engineer in Charge: A History of the Langley Aeronautical Laboratory, 1917- 25 Loftin, Quest for Performance, p. 108. 1958 (Washington, DC: NASA SP-4305, 1987), pp. 198-202. 44 Chapter 3: The Design Revolution On a Streamline to the DC-3 45

be underestimated—not as a valuable type of engineering science, and certainly not has written that “it would have been quite impossible in the prewar period to have in the context of what happened in World War II. By pointing out ways for aircraft any major support from the military, industry, or from Congress for research and to gain a few extra miles per hour, the NACA effort might often have made the development aimed at such radical concepts as the turbojet, the rocket engine, or difference in performance between Allied victory and defeat in the air. In measur- transonic or supersonic aircraft,” and another engineer has commented that “it is able aerodynamic terms, the difference seems miniscule. In the case of the Seversky certain that if the NACA had had the foresight to do research on the turbine engine XP-41, for example, cleanup reduced the airplane’s drag coefficient from 0.0275 to in the decade before World War II, the agency would have met with such technical 0.0226, a difference of 0.0049. But for those who understood the impact of such ridicule and criticism about wasting the taxpayers’ money that it would either have small numbers, this type of difference was whopping. Realizing seemingly tiny dif- had to drop it or have been eliminated.”27 ferences such as this figure provided the coup de grace for the airplane design revo- Luckily, the “failure” of NACA researchers and other American engineers to lution, leading into World War II. anticipate the jet engine’s potential as quickly as a few men in Germany and Great Improved aircraft manufacturing techniques also played a critical role in aero- Britain made little difference in the practical outcome of World War II. A hand- dynamic refinement, particularly through the introduction of flush riveting, which ful of jet planes flew in the war, such as the German Messerschmitt Me 262 and eliminated the dome-shaped rivet heads that protruded untidily above an aircraft’s the British Gloster Meteor, but they really did not fly effectively enough, in great surface (see Document 3-31). Industry even worked to smooth the camouflage enough number, or in well enough synchronization with operating squadrons of paint it used on combat aircraft, trying to minimize the surface roughness of the propeller-driven aircraft, to make a critical difference. Before jet airplanes could paint; this step involved water-sanding the airplane’s surface with fine sandpaper reach their potential, they needed to go through the same sort of exhaustive, sys- and rubbing it down with powdered pumice. So important to the aerodynamic tems-wide development that prop planes experienced in the two decades between performance of an airplane were even the most minor details of design and fabrica- the wars. Jets would eventually go through that years-long, systematic process, as tion that many aircraft manufacturers assembled comprehensive manuals for their we will see documented in a later volume of this overall study. But by the end of the employees by which to ensure an aircraft as “clean” as possible. Document 3-32 pro- war, that process had barely started. vides excerpts from one such manual put together by the Boeing Aircraft Company Propeller-driven aircraft would continue to be built, and continue to develop, in 1940. Of course, under actual combat conditions, it was nearly impossible to for various purposes well after the appearance of successful jets, even to the present. maintain the aircraft in pristine conditions, meaning that the aerodynamic numbers In 1985, Laurence K. Loftin, Jr., who had then recently retired from NASA after achieved in wind tunnel tests were hardly ever achieved in reality. Still, without the a long career in aeronautical research, called the period from 1945 to 1980 one of original “cleaning up,” the airplanes would have turned out to be that much “dirtier” “Design Maturity” (see Document 3-31). In terms of basic configuration, propel- in actual practice. The starting point of the design in terms of aerodynamic clean- ler-driven airplanes would not change much after the war—although some of them ness did make a significant difference. would benefit from the addition of the turboprop, advanced superchargers, and Thus, it was an almost totally reinvented airplane that fought World War II, cabin pressurization. In terms of aerodynamics, however, their level of refinement and no country did more to generate that revolution and then realize its full poten- would never surpass the best aircraft of World War II. Attempts would certainly be tial than the United States, which put the new form of airplane to enormous use in made to apply new ideas in aerodynamics and other areas, and many excellent pro- defeating the Axis powers. Of course, another revolution was already brewing, one peller-driven airplanes would be built, both as transport aircraft, such as the Vickers based on a bold new type of engine, the turbojet. For it, the NACA and the rest of Viscount 810 of the late 1940s and Lockheed Super Constellation and Lockheed C- the American aeronautics community did not provide the visionaries; they surfaced 130 turboprop cargo transport of the 1950s, as well as some remarkable general avi- in Europe, in the rebellious intellects of British engineer Frank Whittle and German ation aircraft, such as the Piper Cherokee, Cessna Skyhawk, Beech Bonanza, Cessna engineer Hans von Ohain. All the while, American aeronautics kept its focus on 310, and the Beech Super King Air. Much more functional and safe “crop dusters” ever-greater refinement of propeller-driven aircraft, a concentration that may have and other agricultural airplanes would fly, such as the Piper Pawnee. Enthusiasts of cost it a chance to explore the concept of jet propulsion more fully—along perhaps “homebuilts” and sailplanes (which, of course, are propellerless) would seek ways with some other revolutionary ideas in aeronautical science and technology such as rockets and cruise missiles. Some critics blamed (and still do blame) the NACA for 27 John V. Becker, High-Speed Frontier: Case Studies of Four NACA Programs, 1920-1950 (Washington, tunnel vision; others defend the organization by saying that there was no way for DC: NASA SP-445, 1980), p. 31, and Ira H. Abbott. “A Review and Commentary of a Thesis it, politically, to get so far out ahead of its clients—and those clients were not inter- by Arthur L. Levine, Entitled ‘A Study of the Major Policy Decisions of the National Advisory ested in pursuing such radical alternative technologies so early. One NACA engineer Committee for Aeronautics,’ Dated 1963,” NASA HQ History Office Archive, HHN-35, Apr. 1964, p. 135. 46 Chapter 3: The Design Revolution of improving aerodynamic performance. But generally speaking, the aerodynamics just did not improve that much. What Loftin observed in terms of “design trends” for propeller-driven aircraft since 1945 was clearly a kind of “steady state” that he predicted would continue indefinitely into the future. When the last volume of The Wind and Beyond is published in a few years, readers will want to compare Loftin’s assessment with that of Dennis Bushnell, also a NASA Langley aerodynamicist. Unlike Loftin, Bushnell will argue in very strong terms in “The Shape of Things to Come – Views of the Future of Aerodynamics,” that aerodynamics has not settled in as a mature or plateau technology and that it has been intellectually constraining to consider it so. Although Bushnell’s historical assessment of what has happened specifically in the design of propeller-driven aircraft will not be much different from Loftin’s, his explanations for why a steady state has prevailed in the second half of the twentieth century will be. In Bushnell’s view, the last word about the form of all aircraft, including those prop-driven, has not yet been written. A new wave of knowledge inspired by a “renaissance of aerodynamics” may change everyone’s opinion about what has happened in the past and what could still happen in the future. Document 3-1 47

The Documents

Document 3-1

Excerpts from Frederick W. Lanchester, Aerodynamics: Constituting the First Volume of a Complete Work on Aerial Flight (New York: D. Van Nostrand Company, 1908), pp. 10-45.

The following excerpts from Frederick W. Lanchester’s 1907 book, Aerodynam- ics, feature his pioneering concept of the “streamline,” one of the central themes of the airplane design revolution that was to unfold in the era between the two world wars. As Lanchester described it in his book, his goal was to define an ideal streamline form (by this, he meant an efficient low-drag shape having ultrasmooth contours offering no resistance to the passage of air along its surface). All of the excerpts below are from chapter one of Lanchester’s book, entitled “Fluid Resistance and its Associated Phenomena.” The author reproduces the continuously running section of the book most concerned with streamlining. Lanchester’s ideas on drag and streamlining directly refuted the widely held but incorrect theory developed earlier by American scientist Samuel P. Langley that skin friction was a negligible factor in aircraft performance. Lanchester was also one of the first to identify and discuss the airflow vortices present at the tips of a wing in flight–a phenomenon that later came to be known as the source for “induced drag.” Lanchester gave only preliminary consideration to the notion of induced drag (which actually proved to be quite significant at low speeds), conjecturing that it was simply a necessary part of the cost of gaining lift. The Englishman’s focus was, rather, on the ideal streamlined shape, one that would minimize other sources of drag besides induced drag.

Document 3-1, Excerpts from Frederick W. Lanchester, Aerodynamics: Constituting the First Volume of a Complete Work on Aerial Flight New York: D. Van Nostrand Company, 1908.

§ 9. On Streamline Form.–When a body of fish-shaped or ichthyoid form travels in the direction of its axis through a frictionless fluid there is no disturbance left in its wake. Now we have seen that in any case the fluid as a whole receives no momentum, so that it is perhaps scarcely legitimate to argue that there is no resistance because there is no communication of momentum, although this is a common 48 Chapter 3: The Design Revolution Document 3-1 49 statement.1 It is clear, however, that if there is no residuary disturbance there is no regard this reaction as the centrifugal component of the curvilinear path of the flow, necessary expenditure of energy, and this equally implies that the resistance is nil. and as such it may be indicated by arrows as in the figure. The fluid in the vicinity of a streamline body is of necessity in a state of motion By assuming the bends in the pipe to be equal and a uniform velocity through- and contains energy, but this energy is conserved, and accompanies the body in its out, it follows that these centrifugal components exactly balance one another, each travels, just as in the case of the energy of a wave. It adds to the kinetic energy of the to each, and the pipe has no unbalanced force tending to push it in one direction body in motion just as would an addition to its mass. or the other. The argument may be found presented in this form in White’s “Naval According to the mathematical theory of Euler and Lagrange, all bodies are of Architecture.” The same net result follows, no matter what the exact form of the streamline form. This conclusion, which would otherwise constitute a reduction ad bends, or whether or not the velocity is uniform, provided the bends are smooth absurdum, is usually explained on the ground that the fluid of theory is inviscid, and the cross-section (and therefore the velocity) is the same at E as at A, for under whereas real fluids possess viscosity. It is questionable whether this explanation these circumstances the pressure at A will be the same as at E, the applied forces alone is adequate. thus being balanced, and there will be no momentum communicated by the fluid in its passage. When a streamline body travels through a fluid the lines of flow may be regarded as passing round it as if conveyed by a number of pipes as in Fig. 2. It is convenient, and it in no way alters the problem, to look upon the body as stationary in an infinite stream of fluid (Fig. 3); we are then able to show clearly the lines of flow relatively to the surface of the body. Now let us take first the fluid stream that skirts the surface itself, and let us suppose this included between the walls of an imaginary pipe, then forces will be developed in a manner represented in Fig. 2, and these forces may be taken as acting on the surface of the body. It is not necessary to suppose that there § 10. Froude’s Demonstration.–An explanation of the manner of the conserva- is actual tension in the fluid, as might be imagined from Fig. 3, where the forces tion of kinetic energy, in the case of a streamline body, has been given by the late act outward from the body, this is obviated by the general hydrostatic pressure that Mr. A.V. Froude. obtains in the region; the forces as drawn are those supplied by the motion of the Referring to Fig. 2, A, B, C, D, E, represents a bent pipe, through which a fluid fluid, and can be looked upon as superposed on those due to the static pressure. is supposed to flow, say in the direction of the lettering, the direction at A and at E being in the same straight line; it is assumed that the fluid is frictionless. Now so long as the bends in the pipe are sufficiently gradual, we know that they cause no sensible resistance to the motion of the fluid. We have excluded viscous resistance by hypothesis, and if the areas at the points A and E are equal there is no change in the kinetic energy. Moreover, the sectional area of the pipe between the points A and E may vary so long as the variations are gradual; change of pressure will accompany change of area on well-known hydrodynamic principles, but no net resistance is introduced; consequently the motion of the fluid through the pipe does not involve any energy expenditure whatever. If, similarly, we deal with the next surrounding layer of fluid, we find that the Let us now examine the forces exerted by the fluid on different portions of pressure to which it gives rise acts to reinforce that of the layer underneath (i.e., the pipe in its passage. The path of the particles of fluid in the length between the nearer the body), and so on, just as in hydrostatics the pressure is continually points A and B is such as denotes upward acceleration, and consequently the fluid increased by the addition of superincumbent layers of fluid, and thus we find that here must be acted on by an upward force supplied by the walls of the pipe, and the the body is subjected to increased pressure acting on its front and rear, and dimin- reaction exerted by the fluid on the pipe is equal and opposite. A shorter way is to ished pressure over its middle portion. Now it has been shown, in the case of the pipe, that the algebraic sum of all forces in the line of motion is zero, so that in the streamline body the sum of the forces produced by the pressure on its surfaces will 1 This somewhat academic objection would cease to apply if any means could be found to properly define the idea which undoubtedly is conveyed to the mind by the argument in question. be zero, that is to say, it will experience no resistance in its motion through the fluid. 50 Chapter 3: The Design Revolution Document 3-1 51

It may be taken as corollary to the above, that in a viscous fluid the resistance pressure measured from the real zero must everywhere be positive, otherwise the of a body of streamline form will be represented approximately by the tangential fluid will become discontinuous and cease to follow the surface. This is a difficulty resistance of its exposed area as determined for a flat plate of the same general pro- that has been actually experienced in connection with screw propellers, and termed portions. This is the form of allowance suggested by Froude; a more elaborate and cavitation. accurate method has been given by Rankine, in which allowance is made for the § 13. The Motion in the Fluid.–It has been shown that the head of a stream- variation in the velocity of the fluid at different points on the surface of the body. line form is surrounded by a region of increased pressure. Consequently the fluid Neither of these methods includes any allowance for viscous loss owing to the dis- as it approaches this region will have its velocity reduced, and the streamlines will tortion of the fluid in the vicinity of the body. widen out, as shown in Fig. 3 (see also Figs. 42, 44, 45, etc.). This behavior of the § 11. The Transference of Energy by the Body.–It is of interest to examine the fluid illustrates a point of considerable importance, which is frequently overlooked. question of the transference of energy through the streamline body itself from one Whenever a body is moving in a fluid, its influence becomes sensible considerably part of the fluid to the other. For the purpose of reference the different portions of in advance of the position it happens to occupy at any instant. The particles of the body have been named as in Fig. 4, the head, the shoulder, the buttock, and fluid commence to adjust themselves to the impending change with just as much the tail, the head and shoulder together being termed (as in naval architecture) the certainty as if the body acted directly on the distant particles through some inde- entrance, and the buttock and tail the run. The dividing line between the entrance pendent agency, and when the body itself arrives on the scene the motion of the and run is situated at the point of maximum section, and the dividing line between fluid is already conformable to its surfaces. There is no impact, as is the case with the head and shoulder on the one hand, and between the buttock and tail on the the Newtonian medium, and the pressure distribution is more often than not quite other, is the line on the surface of the body at which the pressure is that of the different from what might be predicted on the Newtonian basis.–This behavior of hydrostatic “head.” a fluid is due to its continuity. It follows from elementary considerations that the fluid in the “amidships” region possesses a velocity greater than the general velocity of the fluid (the body, as before, being reckoned stationary). We know that at and about the region C, Fig. 3, the fluid has a less area through which to pass than at other points in the field of flow. It is in sum less than the normal area of the stream by the area of cross-section of the body at the point chosen. But the field of flow is made up of a vast number of tubes of flow, so that in general each tube of flow will be contracted to a greater or less extent, the area of section of the tubes being less at points where the area of Now, as the body advances, the head, being subject to pressure in excess of that the body section is greater. We know that a contraction in a tube of flow denotes an due to the hydrostatic “head,” is therefore doing work on the fluid; that is to say, increase of velocity. transmitting energy to the fluid; the shoulder also advancing towards the fluid is Thus on the whole the velocity of the fluid is augmented across any normal subject to pressure less than that due to hydrostatic head, and is consequently receiv- plane that intersects the body itself, but the increase of velocity is not in any sense ing energy from the fluid; the buttock, which is receding from the fluid, is also a uniform in its distribution. In fact, towards the extremities of the body, and in its region of minus pressure and so does work on the fluid; and lastly, the tail is reced- immediate neighborhood, we have already seen that the motion of the fluid is actu- ing under excess pressure and so receives energy. We thus see that there are two ally slower than the general stream. regions, the head and buttock, that give up energy continuously to the fluid, and The motion of the fluid is examined from a quantitative point of view in a two regions, the shoulder and tail, that continuously receive it back again. The con- subsequent chapter (Chap. III.), where plottings are given of the hydrodynamic dition of perfect streamline motion is that the enemy account shall balance. solution in certain cases. § 12. Need for Hydrostatic Pressure, Cavitation.–The motion impressed on the § 14. A Question of Relative Motion.–The motion of the fluid has so far been fluid by the pressure region of the head is compulsory, unless (as may happen in the considered from the point of view of an observer fixed relatively to the body; it will case of a navigable balloon) deformation of the envelope can take place. The motion be found instructive to examine the same motions from the standpoint of the fluid impressed by the shoulder, on the contrary, depends upon hydrostatic pressure, for itself, that is to say, to treat the problem literally as a body moving through the fluid, otherwise there is no obligation on the part of the fluid to follow the surface of instead of as a fluid in motion round a fixed body. the body. Hydrostatic pressure is necessary to prevent the formation of a void. The It is evident that the difference is merely one of relative motion. The problems 52 Chapter 3: The Design Revolution Document 3-1 53

are identical: we require to consider the motions as plotted on co-ordinates belong- the plane a, a, a, and terminate of the surface b, b, b, and the motion will be of the ing to the fluid instead of co-ordinates fixed to the body itself. The relation of the character shown. streamlines (which we have so far discussed) to the paths of motion (which we now The form of the surface b, b, b, will be different for different forms of body. It propose to examine) is analogous to that of the cycloid or trochoid to its generating will evidently approach the plane a, a, a, asymptotically, and generally will tend to circle. form a cusp pointing along the axis of flight. The development of this cusp is great- § 15. Displacement of the Fluid.–An unfamiliar effect of the passage of a body est in cases where the extreme entrance and run are of bluff form, as in the Rankine through a fluid is a permanent displacement of the fluid particles. This displace- Oval, Fig. 42, where the point of the cusp is never reached, the surface approaching ment may be readily demonstrated. If a mass of fluid be moved from any one part of the axis of light asymptotically. In reckoning the displacement of the fluid (§ 15), an enclosure to any other part, the enclosure being supposed filled with fluid, there the volume included in the cusped surface forward of the plane a, a, must be con- is a circulation of fluid from one side to the other during transit; and if we suppose sidered negative, since here the fluid is displaced in the same direction as the motion it to be moved from one side to the other of an imaginary barrier surface, then an of the body. equal volume of fluid must cross the same barrier surface in the opposite direction. § 17. Orbital Motion and Displacement, Experimental Now it is of no importance whether the thing we move be a volume of fluid or a Demonstration.–The displacement of the fluid and the form solid body, so that when a streamline body passes from one side to the other of a of the orbit can be roughly demonstrated by a simple smoke surface composed of adjacent particles of fluid, that surface will undergo displace- experiment. If a smoke cloud be viewed against a dark back- ment in the reverse direction to that in which the body is moving, and the volume ground during the passage of a body of streamline form in its included between the positions occupied before and after transit will be equal to the vicinity, the retrograde movement of the air is clearly visible. volume of the body itself. So long as the surface of the body is not too close, the move- Moreover, since the actual transference of the fluid is clue to a circulation from ment is clean and precise, and the general character of the the advancing to the receding side of the body, it will take place principally in the orbit form can be clearly made out; it is found to be, so far immediate vicinity of the body and less in regions more remote; it is, therefore, as the eye can judge, in complete accord with the foregoing immaterial whether the fluid be contained within an enclosure or whether one or theory. The commencement and end of the orbit, where the more of its confines be free surfaces, provided that continuity is maintained, and motion should be in the same direction as the body, is most that the body is not in the vicinity of a free surface. difficult to observe, though even this detail is visible if the § 16. Orbital Motion of the Fluid Particles.–Since the motion of the fluid results orbit selected be sufficiently near to the axis of flight. The in a permanent displacement, the motion of a particle does not, strictly speaking, difficulty here is that the latter part of the orbit is generally constitute an orbit. It is, however, convenient in cases such as the present to speak lost in consequence of the “frictional wake,”2 i.e., the current of the motion as orbital. set up by viscous stress in the immediate neighborhood of If we could follow the path of a particle along any streamline, and note its the body in motion. In all actual fluids a wake current of this change of position relatively to an imaginary particle moving in the path and with kind is set up, and the displacement surface b, b, b, Fig. 5, the velocity of the undisturbed stream, we should have data for plotting the orbital is obliterated in the neighborhood of its cusp by a region of motion corresponding to the particular streamline chosen. Thus we know that the turbulence. amplitude of the orbit of any particle, measured at right angles to the direction of § 18. Orbital Motion, Rankine’s Investigation.–The form of the orbits of the flight, is equal to that of the corresponding streamline. fluid particles has been investigated theoretically for a certain class of body by Ran- We further know that, in general, the particles have a retrograde motion–that kine (Phil. Trans., 1864). is, their final position is astern of their initial position–also that the maximum ret- Rankine closely studied the streamlines of a body of oval form, generated by a rograde velocity is to be found in the region of maximum amplitude. Beyond this certain method from two foci (§ 77), and by calculation arrived at the equation to we know that the initial motion of any particle is in the same direction as of the the orbit motion of the particles. The result gives a curve whose general appearance body, and that this initial motion is greater for particles near the axis of flight than is given in Fig. 6 (actual plotting), in which the arrows represent the motion of the for those far away. particle, the direction of motion of the body being from left to right. Let b, b, b, etc., Fig. 5, represent the final position of a series of particles originally situated in the plane a, a, a; then the orbits of these particles will originate on 2 A term used in naval architecture. 54 Chapter 3: The Design Revolution Document 3-1 55

Discussing the particular case in which the eccentricity of the oval vanishes, and the whole resistance being then due to the excess pressure region in front of the the form merges into that of a circle, Rankine says,–”…The curvature of the orbit body, the dead-water or wake being at approximately the hydrostatic pressure of varies as the distance of the particle from a line parallel to the axis of X, and midway the fluid. between that axis and the undisturbed position of the particle. This is the property The surface separating the live stream and the dead-water constitutes a disconti- of the looped or coiled elastic curve; therefore when the water-lines are cyclogenous nuity, since the velocity of the fluid, considered as a function of its position in space, the orbit of each particle of water forms one loop of an elastic curve.” Further, he is discontinuous. This case is not one of a physical discontinuity, such as discussed says–”The particle starts from a, is at first pushed forward, then deviates outwards in § 12, for the region on either side of the surface is filled with the same kind of and turns backwards, moving directly against fluid; it is rather a kinetic discontinuity, that is to say a discontinuity of motion. the motion of the solid body as it passes the § 20. The Doctrine of Kinetic Dis- point of greatest breadth, as shown. The par- continuity.–The theory of kinetic discon- ticle then turns inwards, and ends by follow- tinuity is of modern origin, having been ing the body, coming to rest at b in advance introduced and developed by Kirchhoff, of its original position.” Helmholtz, and others, to account for This orbit in some respects resembles the phenomenon of resistance in fluid that arrived at by the author, but differs in motion. The analytical theory, based on the one very important point that, whereas the hypothesis of continuity, does not in the author’s method gives a retrograde dis- general lead to results in harmony with placement of the fluid as the net consequence experience. All bodies, according to the of the passage of the body, Rankine’s conclu- Eulerian theory, are of streamline form, provided that the hydrostatic pressure of the sion is exactly the contrary. fluid is sufficient to prevent cavitation; we know that in practice this is not the case. As the author’s result is capable of experimental verification, it is evident that According to the teaching of Helmholtz and Kirchhoff, a kinetic discontinuity some subtle error must exist in Rankine’s argument, the exact nature of which it is can be treated as if it were a physical discontinuity; that is to say, the contents of the difficult to ascertain. dead-water region can be ignored; and this method of treatment is now generally § 19. Bodies of Imperfect Streamline Form.–In an actual fluid, bodies of other recognized, although not universally so. The controversial aspect of the subject is than streamline form experience resistance apart from that directly due to viscosity. discussed at length at the conclusion of Chap. III. In the practical shaping of a streamline body it is found essential to avoid cor- The principal objection to the theory of discontinuity is that in an inviscid fluid ners or sharp curves in the line of flow. Bodies in which duo precaution is not taken a surface of discontinuity involves rotation, and therefore, by a certain theorem of in this respect offer considerable resistance to motion, and the regions of abrupt Lagrange, it is a condition that cannot be generated.3 A further objection some- curvature give rise to a discontinuity in the motion of the fluid. Thus Fig. 7 repre- times raised is that such a condition as that contemplated would be unstable, and sents a double cone moving axially, and it will be noticed that the flow has not time that the surface of discontinuity, even if formed, would break up into a multitude of to close in round the run, as it would do in a properly formed streamline body, but eddies. Whether this is the case or not in an inviscid fluid, it is certain that in a fluid shoots past the sharp edge, as indicated in the figure. The region in the rear of the possessed of viscosity a surface of discontinuity does commence to break up from body, Z, is filled with fluid that does not partake of the general flow, and which is the instant of its formation; but as this breaking up does not affect the problem in termed dead-water. any important degree, the objection in the case of the inviscid fluid is probably also The resistance experienced by bodies of imperfect form is due to the work done without weight. on the fluid, which is not subsequently given back, as is the case with the streamline In a real fluid a finite difference of velocity on opposite sides of any surface body. This resistance can be traced to two causes, namely, excess pressure on the sur- would betoken an infinite tangential force. Consequently the discontinuity becomes face in presentation and diminished pressure in the dead-water region. The former a stratum rather than a surface, and the stratum will either be a region in which a is of dynamic origin, the energy being expended in directly impressing motion on velocity gradient exists (§ 31), or it will become the seat of turbulent motion (§ 37), portions of the fluid; the latter is due to the entrainment or viscous drag experienced the latter in all probability. by the dead-water at the surface hounded by the live stream. It is generally believed that, in a fluid whose viscosity is negligible, the latter cause would be inoperative, 3 Chap. III. §§ 65–71. 56 Chapter 3: The Design Revolution Document 3-1 57

The conception of the discontinuity as a surface and the method involving The theory of discontinuity this conception are in no way affected by these considerations. The term surface of also receives support of the most discontinuity may be looked upon as an abstraction of that which is essential in a convincing description from the somewhat complex phenomenon. experiments of Mutton, 1788, and § 21. Experimental Demonstration of Kinetic Discontinuity.–The reality and Dines, 1889, by which it is shown importance of the discontinuous type of motion can be demonstrated conclusively that the pressure on a solid hemi- by experiment. sphere, or a hemispherical cup In Fig. 8, a, b, c, is a hollow spherical globe in which d is a tube arranged to (such as used on the Robinson ane- project in the manner shown. An ordinary lamp globe and chimney will be found mometer), both in spherical pre- to answer the purpose the former having one of its apertures closed by a paper disc. sentation, does not differ from that The whole is carefully filled with smoke and then moved through the air in a direc- on a complete sphere to an extent tion from right to left, the relative that experiment will disclose. This direction of the air being indicated not only disposes of the streamline by the arrow. sphere of mathematical concep- It will be found that the air tion, but proves at the same time will enter the tube and displace the approximate constancy of wake the smoke through the annular pressure under variation of rear aperture. The issuing smoke fol- body form. The same lesson is to lows the surface of the sphere in be gleaned from experiments in the the most approved manner as far as case of the hemisphere, cone, and the “equator,” but then passes away circular plate (all in base presenta- at a tangent, the stratum disconti- tion), whose resistance is found to nuity, the dead-water region, and be approximately equal (Fig. 17). the turbulent character of motion, § 22. Wake and Counterwake being all clearly manifest. The dis- Currents.–Reference has already been made to the frictional wake current to which continuity, as may have been anticipated, does not appear as a clean-cut surface; a streamline body gives rise owing to the viscous stress it exerts on the fluid in its it is marked almost from the commencement, as indicated in the figure, by eddy neighborhood. With bodies of imperfect form there is, in addition to the frictional motion; but when we remember that, according to the Fulerian theory, the lines of wake, a wake current constituted by the contents of the dead-water region, that is, flow should carry the smoke along a symmetrical path to the opposite pole of the the fluid contained within the surface of discontinuity. sphere, as in Fig. 45 (Chap. III.), the conclusion is plain. The general motion of the wake current is in the same direction as the body The author has succeeded in photographing the flow round a cylinder in motion itself, but, owing to the viscous drag exerted on it by the surrounding stream, this in a smoke-laden atmosphere (Fig. 9). In this example it may be noticed that the motion has superposed on it one of circulation, which probably results in the cen- surface or stratum of discontinuity arises from a line some distance in front of the tral portion of the wake traveling actually faster than the body4 and the outer part plane of maximum section; the difference in the behavior of a cylinder and sphere slower, though Dines’ experiments seem to point to the disturbance being of so in this respect is clue to the fact that in the former case the lines of flow are cramped complex a character that it is impossible to trace any clearly defined system.5 laterally, the motion being confined to two dimensions, whereas in the 1atter case, Now, since there can be no momentum communicated to the fluid in sum (§ the motion being in three dimensions, the fluid can “get away” with greater facility. 5), there must be surrounding the dead-water or wake current a counter-current This difference is reflected in the lower coefficient of resistance found experimentally for the sphere than that ascertained for the cylinder. Thus in the experiments in the opposite direction to that of the wake, that is, in the reverse direction to the of Dines (§ 226) the pressure per square foot of maximum section on a 5/8-in. cylindrical rod was found to be more than double that on a 6-in. sphere, though 4 Since writing this passage the author has observed this “overtaking” current photographed in Fig. doubtless the difference in size in the bodies compared may contribute something 9. It may be faintly discerned in this Figure in the central region of the “dead-water.” to the disparity. 5 “On Wind Pressure upon an Included Surface,” Proc. Royal Soc., 1980. 58 Chapter 3: The Design Revolution Document 3-1 59

motion of the body; and this counterwake current is being continuously generated, nuity will be developed, as indicated in the figure. Let now the detached portion just as the wake current itself, and contains momentum equal and opposite to that be replaced. Then the question arises, What are the changed conditions that will of the wake. When in a fluid possessing viscosity the wake and counterwake cur- interfere with the permanence of the discontinuous system of flow, as depicted in rents intermingle by virtue of the viscous connection between them, and become the figure? involved in a general turbulence, the plus and minus momenta mutually cancel, and If, in the first place, the fluid be taken as inviscid, and if, for the purpose of the final condition of the fluid at all points is one of zero momentum. argument, we assume that the system of flow indicated in the figure is possible in We may regard the counterwake current as a survival of the motion which, we an inviscid fluid, then it is evident that when the run is replaced we shall not have have shown, must exist in the neighborhood of the maximum section of a stream- disturbed the conditions of flow, for our operations have been confined to the dead line body (§ 13) opposite in direction to its motion through the fluid. The failure of water region, where the fluid is at rest relatively to the body. Consequently the dis- the stream to close in behind the body means that this motion will persist. continuous system of flow will persist. That is to say, under the supposed conditions The mingling of the wake and counterwake may be regarded as a phenomenon streamline motion is either unstable or is at best a condition of neutral equilibrium. quite apart from the initial disturbance, and the turbulence or otherwise of the wake Let us next introduce viscosity as a factor. The conditions are now altered, for the does not materially add to or detract from the pressure on the front face of the body, fluid in the dead-water region is but concerns merely the ultimate disposal of the energy left behind in the fluid. no longer motionless, but is in No distinction is necessary between the frictional wake and the dead-water active circulation, and the intro- wake so far as the production of a counterwake current is concerned. The total duction of the rear half of the wake current is the sum of the two, and the total counterwake is equal and opposite body obstructs the free path of to the total wake. the fluid, so that, as the outer § 23. Streamline Motion in the Light of the Theory of Discontinuity.–The layers of the dead-water are car- theory of kinetic discontinuity presents the subject of streamline motion in a new ried away by the viscous drag, the light, and enables us to formulate a true definition of streamline form. Thus– fluid in the interior has difficulty A streamline body is one that in its motion through a fluid does not give rise to in finding its way back to take its place. This difficulty is greatest in the region from a surface of discontinuity. which the discontinuity springs, where the dead-water runs off to a “feather edge,” In the previous discussion, § 9 et seq., no attempt has been made to delineate and it is evident that some point of attenuation is reached at which the return flow streamline form, that is to say (according to the present definition), the form of becomes impossible, and the fluid will be “pumped out” or ejected from the region body that in its motion through a fluid will not give rise to discontinuity. It has been forward of this point. This brings the discontinuity further aft on the body, where assumed that such a body is a possibility, and from the physical requirements of the the process can be supposed repeated, so that eventually the whole dead-water has case the general character of the body form has been taken for granted. been pumped away, and streamline motion supervenes. It is evident that the process Under our definition, if, as in the mathematical (Eulerian) theory, we assume will not occur in stages, as above suggested, but will be continuous. continuity as hypothesis, then all bodies must be streamline, which is the well- It might be supposed from the foregoing argument that the degree of curvature known consequence. If, on the other hand, as in the Newtonian medium, we of the surface of the body would not be a matter of importance, as in any case the assume discontinuity, then it is evident by our definition that streamline form can feather edge of the dead-water would be sufficiently fine to ensure the ejection of have no existence, which, again, is what we know to be the case. It remains for us to some small amount of the fluid, and this process by continuous repetition would demonstrate, on the assumption of the properties of an ordinary fluid, the condi- eventually clear the wake of its contents. If the surface of the body were frictionless, tions which govern the existence or otherwise of discontinuity, and so control the doubtless this might be the case, but it is established that there is continuity between form of a streamline body. the surface of an immersed body and the surrounding fluid; that is to say, there is In order that streamline motion should be possible such motion must be a the same degree of viscous connection between the fluid and the surface as there is stable state, so that, if we suppose that by some means a surface of discontinuity between one layer of the fluid and another. The consequence of this is that the dead- be initiated, the conditions must be such that the form of motion so produced is water never fines off entirely, but extends forward as a sort of sheath enveloping the unstable. whole surface of the body, and if the curvature at any point is too rapid, the ejection Let us suppose that we have (Fig. 10) a streamline body made in two halves, may not prove effective, and the discontinuity will persist. It is evident therefore and that the rear half, or run, be temporarily removed; then a surface of disconti- that there will be some relation between the bluffness of form permissible and the 60 Chapter 3: The Design Revolution Document 3-1 61 viscosity of the fluid, and, other things being equal, the less the viscosity the finer this type of vessel is one of which will have to be the lines of the body. The theory evidently also points to the impor- but little information has been tance of smoothness of surface when the critical conditions are approached. published. The subject is not yet exhausted. We know that the thickness of the stratum In Figs. 11 and 12 curves of fluid infected by skin friction increases with the distance from the “cut-water”; are given whose ordinates repre- that is to say, the factor on which the curvature of the surface probably depends sent the area of cross-section at is relatively more important on the buttock than on the shoulder. Hence we may different points. This curve has expect that the lines of entrance can with impunity be made less fine than the lines been obtained by differentiating a of the run. displacement curve plotted from Again, all forces due to the inertia of the fluid vary as the square of the velocity; a series of immersion measure- those due to viscosity vary in the direct ratio of the velocity (§ 31). Therefore for dif- ments. These measurements were ferent velocities the influence of viscosity predominates for low velocities, and that made by a method of displace- of inertia when the velocity is high. Consequently the form suited to high velocity ment, the fish, suspended tail will be that appropriate to low viscosity, and vice versa; that is to say, the higher the downward, being lowered stage velocity the finer will be the lines required. by stage into a vessel of water, § 24. Streamline Form in Practice.–The practical aspect of streamline form may measurements being made of the be best studied from the bodies of fishes and birds, the lines of which have been overflow. gradually evolved by nature to meet the requirements of least resistance for motion The area curves have been further translated into the form of solids of revolu- through a fluid, water or air, as the case may be. tion, which may be taken as the equivalent of the original form in each case. Some Since all animals have func- doubt exists as to the exact form in the region of the head, owing to the water enter- tions to perform other than mere ing the gills. The effect of this is very evident in the case of the trout (Fig. 12), where locomotion, we find great diver- the form has been “made good” by a dotted line. sity of detail, and we frequently For the purpose of meet with features whose exis- comparison outline eleva- tence is in no way connected tions are given in Fig. 14 of with the present subject. We three types of Whitehead may readily recognize in these torpedo. These are forms cases the exceptional develop- that have been developed ment of certain organs or parts by long experience, but the to meet the special requirements shape is largely dictated by of a particular species, and by a special considerations. The sufficiently wide selection we bluff form of head, for example, in models A and C is adopted in order to bring the can eliminate features that are explosive charge into as close proximity as possible to the object attacked. It prob- not common, and so arrive at ably also gives a form that is more easily steered. an appreciation of that which is § 25. Streamline Form.–Theory and Practice Compared.–Before a rigid com- essential. Thus the herring (Fig. 11), the trout (Fig. 12), or the salmon (Fig. 13) may parison can be instituted between the theoretical results of § 23 and the actual forms be cited its typically fish shaped fish. found in nature considerable further information is required. We do not know with Beyond the lessons to be derived from these natural forms, there is very little accuracy the speeds for which the different fish forms have been designed or are best practical information available. The lines of ships are governed by considerations adapted. We also lack knowledge on certain other important points. The present foreign to the subject, the question of wave-malting, for example, being a matter of comparison must therefore be confined to generalities. vital importance. The submarine has not yet reached its stage of development that In the first place, we may take it that the conclusion as to the bluffer form being would justify its form being taken as a fully evolved model; also, for obvious reasons, that suited to greater viscosity is fully borne out in practice, though the whole of the 62 Chapter 3: The Design Revolution Document 3-1 63

considerations bearing on this point are not here available. It is explained in Chap. ously expended. This performance of work is otherwise represented by a resistance II. that the viscosity divided by density (or kinetic viscosity) is the proper criterion to motion, being the difference between the excess pressure on the head and the in such a case as that under discussion, and on this basis air is far more viscous than diminished pressure on the shoulder, according to the principle explained in § 11. water, so that we shall expect to find aerial forms bluffer in their lines than aqueous If now we restore the buttock, so that the mutilation is confined to the simple loss of the tail (Fig. 15), the diminished pressure on the buttock acts as a drag upon the body, and more work must be expended in propulsion. This additional energy will appear in the fluid as a radial component in the motion of the stream which does not exist if the whole run is removed. It is probable that some of this energy is restored by an increase in the pressure of the dead water due to the converging stream, but we have no means of making a quantitative computation. An illustration of this principle may be cited in the type of hull employed in a modern racing launch. The stern is cut off square and clean, and may constitute the maximum immersed section. There would seem in fact to be no logical compromise between a boat with an ordinary well-proportioned entrance and run, and one in which the latter is sacrificed entirely. In such a form, when traveling at high speed the water quits the transom entirely, and consequently sacrifice is made of the hydrostatic pressure on the immersed transom area. The point at which the front half of a boat thus takes less power for its propulsion than the whole is probably about that speed at which the skin friction on the run (the after-half), if present, exceeds the hydrostatic pressure on the maximum immersed section. This does not, however, determine the point at forms. Taking the solid of revolution as the basis of comparison, we have in the case which it pays to make the sacri- of the herring and the trout the length approximately seven times the maximum fice, owing to the fact that for the diameter. The general ratio found amongst bird forms is about three or four to same capacity the truncated form one, the samples chosen for measurement being as far apart as the albatross and the has to be that of a larger model. common sparrow. Consequently we find that the theoretical conclusion receives The rating rule also exerts an arbi- substantial confirmation. trary influence. When, as is usual, The relation of fineness to speed is not so easy of demonstration, owing to the the length is penalized, an addi- absence of accurate data. It would, however, seem to be sufficiently obvious as a tional inducement is offered for matter of general experience that our conclusions hold good. It is almost certain that the designer to adopt the trun- in general the fish with the finer lines are the faster swimmers. If this conclusion be cated type. When the truncated type of hull is adopted it is advantageous to employ accepted, the viscosity relation of the preceding paragraph is emphasized, for there is shallow draught, for the hydrostatic pressure for a given displacement is less. This no doubt that the average speed of flight is greatly in excess of any ordinary velocity form is also partly dictated by considerations relating to propulsion. attained by fish. § 27. Mutilation of the Streamline Form (continued).–In Fig. 16, A and B, the § 26. Mutilation of the Streamline Form.–There are certain types of body that consequences of truncating the fore body, or entrance, of a streamline body are indi- may be regarded as mutilations of the streamline form, and the consequences of cated diagrammatically. If, as in A, the mutilation be slight, the result may be merely such mutilation may now be examined. a local disturbance of the lines of flow. A surface of discontinuity will probably arise, If, in the case of a body propelled at a constant velocity, the entire run be originating and terminating on the surface of the body in the manner shown. It is removed, as in § 23, the consequence is a surface of discontinuity emanating from possible that if the streamline body be traveling at something approaching its criti- the periphery of section. Under these circumstances, if we neglect the influence of cal velocity (at which even in its complete form it is on the point of giving rise to viscosity and the consequent loss of wake pressure, the work done appears wholly discontinuity), a minor mutilation such as here suggested might have more serious in the counterwake current, on the production of which energy is being continu- consequences. 64 Chapter 3: The Design Revolution Document 3-1 65

If the greater part of the lines of flow of characteristic streamline form. We may therefore generalize and say, entrance be removed, as shown at All bodies passing through a fluid are surrounded by a streamline system of a flow B, the surface of discontinuity gen- of a greater of less degree of perfection depending upon the conformability or other- erated quits the body for good, and wise of the surface of surfaces of the body. the resistance becomes immediately This proposition, if not sufficiently as great as that of a normal plane of obvious from the considerations above area and form equal to that of the given, may easily be demonstrated experi- section. This is in harmony with the mentally. experiments of Hutton and Dines, In the experiment described in § 17, the orbital motion of the particles of the to which reference has already been fluid is demonstrated by the motion of an ichthyoid body in air irregularly charged made (Fig. 17), the three bodies with smoke. This orbital motion, with its consequent displacement, is quite charac- shown being found to offer the teristic, and if other shapes of the body be substituted for the streamline form, the same resistance within the limits of motion can be observed much closer to the axis of flight than is the case for a sphere experimental error. or other bluff form; also when the movement is complete nothing further happens. It is evident that the dictum of the late Mr. Froude, that it is “blunt tails rather In the case of a sphere, the looked for movement duly takes place; but immediately than blunt noses that cause eddies” (and therefore involve a loss of power), is appli- after the whole of the fluid under observation is involved in a state of seething cable only to bodies having already some approximation to streamline form. It is turbulence, where the wake and counterwake currents are mingling. If the point of obviously useless to provide a nice sharp tail if previous attention has not been observation is sufficiently remote, the orbital motion may be detected, even in the given to the shoulder and buttock lines. Mr. Froude probably meant that in a well- case of the normal plane, beyond the immediate reach of the wake turbulence. designed streamline form the tail should be finer in form than the head, a matter § 29. Displacement due to Fluid in Motion.–It has been shown (§ 15) that that up to his time had presumably been neglected. the fluid in the neighborhood of the path of flight of a streamline body under- The primary importance of easy shoulder lines has been long recognized as a goes displacement, and that the total fundamental feature in the design of projectiles. A full-sized section of a Metford displacement is equal to the volume 303 bullet, illustrating this point, is given in Fig. 18, and a streamline form of which of the body. It might be expected in it may be regarded as a “mutilation” is indicated by the dotted line. the case of the normal plane, which § 28. Streamline Flow General.–Let us sup- possesses no volume, that the dispose an approximate streamline form to be built of placement would be nil, and such bricks, and, in the first place, we will assume that the would doubtless be the case if the bricks are so small as to merely give rise to a superfi- form of flow were that of the Eule- cial roughness. Then this roughness will add to the rian theory. skin friction and will give rise to some local turbu- In actuality the normal plane, in common with bodies of bluff form, carries a lence, but the general character of the flow system quantity of fluid bodily in its wake, which from the present point of view becomes remains as before. We may go further and suppose in effect part of the body, so that the displacement manifests itself just as if the plane the bricks so large as to form steps capable of giving were possessed of volume. This is characteristic of all bodies that give rise to dis- rise to surfaces of discontinuity (Fig. 19). Then the continuous motion; the displacement is greater than the actual volume of the body. resistance will be increased, and the layer of fluid If there were no mingling of the wake and counterwake currents, the displacement next the body will be violently stirred up; but if we would be infinite, for the counterwake current would persist indefinitely. examine the fluid some distance away we shall still In the case of a streamline body, a certain amount of fluid is carried along find it comparatively unaffected. If we now suppose with the body by viscosity, and this similarly increases the effective displacement the body to consist of a few large blocks, the depth volume. of fluid affected by turbulence will be greater, but at It would appear from actual observation that, where the displacement is due a sufficient distance away we may still expect to find to the attendant fluid, the outer streamlines have a motion closely resembling that 66 Chapter 3: The Design Revolution Document 3-1 67

produced by a streamline body, but that those nearer the axis of flight terminate in The actual means by which the reac- the turbulent wake; the commencement of the orbit is all that can be seen. tion acting on the ball comes about may § 30. Examples Illustrating Effects of Discontinuous Motion.–On the practical be understood from either of two points importance of the study of motion of the discontinuous type it is unnecessary to of view. We may (Fig. 22) regard this dwell. It is at present the only basis on which it is possible to account for the phe- reaction as the centrifugal effect of the nomenon of fluid resistance as experimentally known. Beyond this there are many air passing over the ball preponderat- examples and illustrations which are of especial interest, considered either as proofs ing greatly over that of the fluid passing of the theory itself or in relation to their actual consequence or utility. underneath, or if we anticipate a knowl- A useful application of the principle is found in the screen employed on fast edge of hydrodynamic theory (Chap. steamships to protect the navigating officer, and frequently the “watch,” from the III.), we know that the greater proximity rush of air, without obstructing the field of vision. This is illustrated diagrammati- of the lines of flow in the former region cally in Fig. 20, in which it will be seen that the live stream is carried clear over the is alone sufficient to indicate diminished sailor’s head, the latter being protected by the surface of discontinuity. A similar pressure. The lines as drawn in the figure device is frequently adopted in connection with the dashboard of a motor car. are not plottings–there is no way known Evidence of the most striking kind of of plotting a field of flow of this degree the existence of a surface of discontinu- of complexity–but they may be taken as a very fair representation of what the plot- ity is sometimes met with in the growth ting would be if it could be effected. of trees in the immediate vicinity of the The reason that the streamlines have been shown rising to meet the ball in its edge of a cliff (Fig. 21). It may be seen progress will be better understood in the light of Chaps. III. and IV. This detail is that the form of the surface is clearly related to more advanced considerations than can be entered into at present. delineated, the tree top being cut away as A further interesting example is found in the aerial tourbillion6 (Fig. 23), in though it might have been sheared off by which the rotor K is a stick of segmental section mounted to revolve freely about the a stroke of a mighty scythe. axis L. The plane face of the rotor is set truly An interesting example of an indirect at right angles to the axis of rotation. If this effect of discontinuity is to be found in apparatus be held in a current of air with the the effect of “cut” or “side” on the flight plane face fronting the wind, as, for instance, of a ball. Let a ball (Fig. 22) moving in by holding it outside the window of a rail- the direction of the arrow A have a spin way carriage in motion, the rotor evinces no in the direction of the arrow B. Now tendency to go round in the one direction where the direction of motion of the sur- or the other. If, however, a considerable ini- face of the ball is the same as the relative tial spin be imparted in either direction, the motion of the fluid, as at D, the surface will assist the stream in ejecting the dead wind will suddenly get a bite, so to speak, water, so that the discontinuity will be delayed, and will only make its appearance at and the rotor will gather speed and spin at a point some distance further aft than usual. On the other hand, on the side that is an enormous rate, as if it were furnished opposing the stream the surface of the ball will pump air in, and so assist the discon- with sails like a well-designed windmill. tinuity, which will make its appearance prematurely. The net result of this is that the Referring to Fig. 24, we have at a the type of flow illustrated to which the blade counterwake will have a lateral component (downwards in the figure), and, on the of the rotor will give rise when its motion is normal to the air; b similarly indi- principle of the continuous communication of momentum, there will be a reaction on the ball in the opposite direction, that is to say upwards. A ball may therefore be sustained against gravity or be made to “soar” by receiving a spin in the direction 6 This interesting aerodynamic puzzle was first brought to the notice of the author by .Mr Henry Lea, shown, or, if the spin be about a vertical axis, the path of the ball will be a curve (in consulting engineer, of Birmingham, who, it would appear, had it communicated to him by Mr. A. plan), such that the aerodynamic reaction will be balanced by centrifugal force. S. Dixon, who in turn had it show him when traveling in Italy by Mr. Patrick Alexander. The author has taken no steps to trace the matter further. The explanation here given is his own. 68 Chapter 3: The Design Revolution

cates the form of flow when the rotor is going round slowly, not fast enough for the air to take hold. In both these figures we have the flow independent of the “rear body form,” and the rotor behaves just as if it were a flat plate. Now, let us suppose that the rotor be given a sufficient initial spin to bring about the state of things represented at c. The surface of discontinuity that ordinarily springs from the leading edge has got so close to the rear body of the rotor as to have ejected the “dead- water” on that side, and the resulting form of flow will be something like that illustrated in Fig. 25. Here the pressure on the left-hand side (as shown) will be that of the “dead-water,” which is, as we know, somewhat less than that of hydrostatic head, while that on the right hand side will, owing to the centrifugal component of the stream, be very

much lower; that is to say, the rotor will experience a force acting from left to right which is in the direction of the initial spin, so that the motion will be accelerated and will continue. The fact that the propel- ling force only comes into existence when the initial spin is sufficient to eject the dead water from the leading aide of the rotor blade fully explains the observed fact that a very considerable initial spin is necessary. Document 3-2 69

Document 3-2

Ludwig Prandtl, “Motion of Fluids with Very Little Viscosity,” NACA Technical Memorandum No. 452 (Washington, DC, 1928), translation of “Über Flüssigkeitsbewegung bei sehr kleiner Reibung,” in Vier Abhandlungen zur Hydrodynamik und Aerodynamik, ed. L. Prandtl and A. Betz (Göttingen: Selbstverlag der AVA, 1927, reissued 1944), pp. 1-8. The 1927 publication was a reprint of Prandtl’s 1904 paper that had appeared in Verhandlungen des III. Internationalen Mathematiker-Kongresses, Heidelberg, 1904, ed. A. Krazer (Leipzig: Teubner, 1905), pp. 484-91.

This paper in which Germany’s Dr. Ludwig Prandtl of the University of Göt- tingen introduced the concept of what came to be known as the “boundary layer” is, without question, one of the most classic expressions in the entire history of aerodynamics. Never before had anyone in fluid mechanics so embodied the ideal of bind- ing theory and practice into a unified whole, one in which abstract theorems and experimental facts worked together for the purpose of fundamental applications. In his 1982 book, Bringing Aerodynamics to America, historian Paul A. Hanle examined “The Achievement of Ludwig Prandtl” in detail. As Hanle explained, Prandtl’s was an achievement crucial to “bringing aerodynamics to America” notably because two of his prize students, Max Munk and Theodore von Kármán, were to play critical roles after 1920 in the development of aeronautical research and development (R & D) capabilities in the United States. In Hanle’s view, Prandtl’s 1904 paper typified not only his own mode of research but also that of coming generations of scientists and engineers dedicated to aerodynamic R & D. Prandtl’s arguments were “essentially physical,” based on his “intuition reinforced by his con- firming experiments.” He gave roughly equal time to explaining his theory and to verifying it by reference to numerous drawings and photographs of flow taken from experiments in a water canal. The 1904 paper presented an approximate solution to the problem he posed, arrived at by numerical computation. Although highly technical, the piece could be easily surveyed. Even today, it deserves our attention for what it reveals about Prandtl’s scientific turn of mind, an approach to solving problems that would influence American aerodynamics in many essential ways for years to come (Bringing Aerodynamics to America [Cambridge, MA, and London: The MIT Press, 1982], pp. 43-4.). The central idea of Prandtl’s paper rested in two sentences stated right near the beginning: “I have set myself the task of investigating systematically the motion 70 Chapter 3: The Design Revolution Document 3-2 71

of a fluid of which the internal resistance can be assumed very small. In fact, the Document 3-2, Ludwig Prandtl, “Motion of Fluids with Very Little Viscosity,” resistance is supposed to be so small that it can be neglected wherever great velocity NACA Technical Memorandum No. 452 Washington, DC, 1928. differences or cumulative effects of the resistance do not exist.” As Hanle explained, what Prandtl called his “systematic investigation” in truth consisted of “treating no TECHNICAL MEMORANDUMS more than a few ‘single questions’ in broad outline” (Hanle, pp. 44-5). From this basis, it did not take him long to present the most important point of his paper: “By NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS far, the most important question of this problem is the behavior of the fluid at the surfaces of the solid body.” Sensing something physical about the airflow, Prandtl postulated the existence of a thin “transition layer” in which viscous effects were in fact extremely significant. Two insights in particular led him to hypothesize what came to be known as the boundary later; they were, in Hanle’s words, that “the NO. 452 fluid must be immobile at the surface of the body” and that “the fluid must follow classical hydrodynamic streamlines only a short distance from the surface” (Hanle, p. 45). In laying out a plan for solving problems involving the transition layer and presenting the necessary governing equations, Prandtl made the consequences of his hypothesis clear: “According to the above, the treatment of a certain process of MOTION OF FLUIDS WITH VERY LITTLE VISCOSITY flow therefore reduces to two parts in mutual interaction: one has on the one hand a free fluid, which can be treated as inviscid according to the vortex principles of By L. Prandtl Helmholtz; on the other hand, the transition layers at the fixed boundaries, whose motion is governed by the free fluid, but which for their part give the free motion From “Vier Abhandlungen zur Hydrodynamik and Aerodynamik” its characteristic features by the emission of vortex layers.” Three other points should be made about this paper (all of them echoing Hanle) Gottingen, 1927 before sending the reader off to discover Prandtl’s astuteness up close and personally. First, Prandtl gave absolutely no thought at this point in 1904 to the relevance of his hypothesis to the science of flight. He did not even become slightly interested in any theory of flight until 1906. So, “rather than his interest in flight leading Prandtl to the discovery of the boundary layer, cause and effect were likely reversed” (Hanle, Washington p. 47). March, 1928 Second, as suggested earlier, Prandlt’s paper made little impression on the 80 or so people attending the Third International Congress of Mathematicians at Hei- delberg. This helps to explain why it took as long as it did for his boundary-layer hypothesis to capture any widespread attention. Finally, it is remarkable, given its venerated place in the history of aerodynamics, that the 1904 paper was only eight pages long. Nevertheless, in those few pages, Prandtl laid the foundation for the modern theory of aerodynamic drag. In doing so, he provided in a sort of time capsule an important building block for the airplane design revolution of the 1920s and 1930s. 72 Chapter 3: The Design Revolution Document 3-2 73

National Advisory Committee For Aeronautics Technical term v o ∆ v (change of velocity due to change of location) is large enough to let the Memorandum No. 452 Motion Of Fluids With Very Little viscosity effect appear quite subordinate. The latter almost always happens in cases of fluid motion occurring in technology. It is therefore logical simply to use here Viscosity.* the equation for non-viscous fluids. It is known, however, that the solutions of this equation generally agree very poorly with experience. I will recall only the Dirichlet By L. Prandtl. sphere, which, according to the theory, should move without friction. I have now set myself the task to investigate systematically the laws of motion In classic hydrodynamics the motion of nonviscous fluids is chiefly discussed. of a fluid whose viscosity is assumed to be very small. The viscosity is supposed to For the motion of viscous fluids, we have the differential equation whose evalua- be so small that it can be disregarded wherever there are no great velocity differences tion has been well confirmed by physical observations. As for solutions of this dif- nor accumulative effects. This plan has proved to be very fruitful, in that, on the ferential equation, we have, aside from unidimensional problems like those given one hand, it produces mathematical formulas, which enable a solution of the prob- by Lord Rayleigh (Proceedings of the London Mathematical Society, 11 page 57 = lems and, on the other hand, the agreement with observations promises to be very Papers I page 474 ff.), only the ones in which the inertia of the fluid is disregarded satisfactory. To mention one instance now: when, for example, in the steady motion or plays no important role. The bidimensional and tridimensional problems, taking around a sphere, there is a transition from the motion with viscosity to the limit of viscosity and inertia into account, still await solution. This is probably due to the nonviscosity, then something quite different from the Dirichlet motion is produced. troublesome properties of the differential equation. In the “Vector Symbolics” of The latter is then only an initial condition, which is soon disturbed by the effect of Gibbs, ** this reads an ever-so-small viscosity. I will not take up the individual problems. The force on the unit area, due to ρ((∂/v)/(∂/t) + v ∆v) + ∆ (V + p) = k∆2v the viscosity, is 0 K = k ∆2 v in which v is the velocity; ρ, the density; V, a function of the power; p, pressure; k, viscosity constant. There is also the continuity equation If the vortex is represented by w = ½ rot v, then K = −2 k rot w, according to a well-known vector analytical transformation, taking into consideration that div v = div v = 0 0. From this it follows directly that, for w = 0, also K = 0, that is, that however great the viscosity, a vortexless flow is possible. If, however, this is not obtained in certain for incompressible fluids, which alone will be here considered. cases, it is due to the fact that turbulent fluid from the boundary is injected into the vortexless flow. With a periodic or cyclic motion, the effect of viscosity, even when it is very small, can accumulate with time. For permanence, therefore, the work of K, that is, * “Ueber Flussigkeitsbewegung bei sehr kleiner Reibung.” This paper was read the line integral ∫ K o d s along every streamline with cyclic motions, must be zero before the Third International Congress of Mathematicians at Heidelberg in 1904. for a full cycle. From “Vier Abhandlungen zur Hydrodynamik und Aerodynamik,” pp. 1-8, Got- ∫ K o d s = (V2 + p2) − (V1 + p1). tingen, 1927. A general formula for the distribution of the vortex can be derived from this ** aob scalar product, a x b vector product, ∆ Hamilton differentiator with the aid of the Helmholtz vortex laws for bidimensional motions which have a flow function ψ (Cf. “Encyklopadie der mathematischen Wissenschaften,” Vol. IV, (∆ = i (∂/∂x) + j (∂/∂y) + k (∂/∂z). 14, 7). With steady flow we obtain * From the differential equation, it is easy to infer that, for sufficiently slow and ψ also slowly changing motions, the factor ρ, in contrast with the other time, can −(dw/d ) = (V2 + p2) − (V1 + p1) / P be as small as desired, so that the effect of the inertia can here be disregarded with 2 k ∫ v o d s sufficient approximation. Conversely, with sufficiently rapid motion, the quadratic 74 Chapter 3: The Design Revolution Document 3-2 75

With closed streamlines this becomes zero. Hence we obtain the simple result methods (Cf. Kutta, “Zeitschrift fur Math. und Physik,” Vol. 46, p. 435). One dif- that, within a region of closed streamlines, the vortex assumes a constant value. For ficulty, however, consists in the various singularities developed on the solid surface. axially symmetrical motions with the flow in meridian planes, the vortex for closed The simplest case of the conditions here considered is when the water flows along streamlines is proportional to the radius w = cr. This gives a force K = 4 kc in the a flat thin plate. Here a reduction of the variables is possible and we can write u = direction of the axis. f(y/√x). By the numerical solution of the resulting differential equation, we obtain The most important aspect of the problem is the behavior of the fluid on the for the drag the formula surface of the solid body. Sufficient account can be taken of the physical phenomena 3 in the boundary layer between the fluid and the solid body by assuming that the R = 1.1 … b√( k ρ l u0 fluid adheres to the surface and that, therefore, the velocity is either zero or equal to

the velocity of the body. If, however, the viscosity is very slight and the path of the (b width, l length of plate, u0 velocity of undisturbed water opposite plate). flow along the surface is not too long, then the velocity will have its normal value Figure 1 shows the course of u. in immediate proximity to the surface. The most important practical result of these investigations is that, in certain In the thin transition layer, the great cases, the flow separates from the surface at a point entirely determined by external velocity differences will then produce conditions (Fig. 2). A fluid layer, which is set in rotation by the friction on the wall, noticeable effects in spite of the small is thus forced into the free fluid and, in accomplishing a complete transformation of viscosity constants. the flow, plays the same role as the Helmholtz separation layers. A change in the vis- This problem can be handled best cosity constants k simply changes the thickness of the turbulent layer (proportional by systematic omissions in the general differential equation. If k is taken as small in the second order, then the thickness of the transition layer will be small in the first order, like the normal components of the velocity. The lateral pressure differences can be disregarded, as likewise any curvature of the streamlines. The pressure distribution will be impressed on the transition layer by the free fluid. For the problem which has thus far been discussed, we obtain in the steady condition (X-direction tangential, Y-direction normal, u and v the corresponding velocity components) the differential equation

ρ (u (∂u/∂x) + v (∂u/∂y) + dp/dx = k (∂2u/∂y2) and ∂u/∂x + ∂v/∂y = 0. to the quantity √(kl/ρu)), everything else remaining unchanged. It is therefore pos- * According to Helmholtz, the vortex of a particle is permanently proportional sible to pass to the limit k = 0 and still retain the same flow figure. to its length in the direction of the vortex axis. Hence we have, with steady even flow As shown by closer consideration, the necessary condition for the separation of on each streamline (ψ = const.), w constant, consequently w = fψ. Herewith the flow is that there should be a pressure increase along the surface in the direction of the flow. The necessary magnitude of this pressure increase in definite cases ∫ K o d s = 2k∫ rot w o d s = 2k f’ (ψ) ∫ rot ψ o ds = 2k f’ (ψ) ∫ v o d s. can be determined only by the numerical evaluation of the problem which is yet to be undertaken. As a plausible reason for the separation of the flow, it may be If, as usual, dp/dx is given throughout, as also the course of u for the initial cross stated that, with a pressure increase, the free fluid, its kinetic energy is partially con- section, then every numerical problem of this kind can be numerically solved, by verted into potential energy. The transition layers, however, have lost a large part of obtaining the corresponding ∂u/∂x by squaring every u. Thus we can always make their kinetic energy and no longer possess enough energy to penetrate the region of progress in the X-direction with the aid of one of the well-known approximation higher pressure. They are therefore deflected laterally. 76 Chapter 3: The Design Revolution Document 3-2 77

According to the preced- the upper channel comparatively ing, the treatment of a given free from vortices, the object to flow process is resolved into two be tested being introduced at c. components mutually related to Fine scales of micaceous iron one another. On the one hand, ore are suspended in the water. we have the free fluid, which These scales indicate the nature can be treated as nonviscous of the flow, especially as regards according to the Helmholtz the vortices, by the peculiarities vortex laws, while, on the other of their reflection due to their hand, we have the transition orientation. layers on the solid boundaries, The accompanying pho- whose motion is determined by the free fluid, but which, in their turn, impart their tographs were obtained in this characteristic impress to the free flow by the emission of turbulent layers. manner, the flow being from left I have attempted, in a few cases, to illustrate the process more clearly by dia- to right. Nos. 1-4 show the flow past a wall projecting into the current. The separat- grams of the streamlines, though no claim is made to quantitative accuracy. In so far ing or boundary layer, which passes off from the edge, is apparent. In No. 1 it is very as the flow is vortex-free, one can, in drawing, take advantage of the circumstance, small; in No. 2, concealed by strong disturbances; in No. 3, the vortex spreads over that the streamlines form a quadratic system of curves with the lines of constant the whole picture; in No. 4, the permanent condition is shown. A disturbance is potential. also evident above the wall. Since a higher pressure prevails in the corner, due to the Figures 3-4 show, in two stages, obstruction of the water flow, even here the flow separates from the wall after awhile the beginning of the flow around (Cf. Figs. 1-4). The various striae a wall projecting into the current. visible in the vortex-free portion The vortex-free initial flow is rap- of the flow (especially in Nos. 1- idly transformed by a spiral sepa- 2) are due to the fact that, at the rating layer. The vortex continually inception of the flow, the liquid advances, leaving still water behind was not entirely quiet. Nos. 5-6 the finally stationary separating show the flow around a curved layer. obstacle or, from another view- Figures 5-6 illustrate the analo- point, through a continuously gous process with a cylinder. The narrowing and then widening fluid layers set in rotation by the channel. No. 5 was taken shortly friction are plainly indicated. Here after the inception of the flow. also the separating layers extend One boundary layer has devel- into infinity. All these separating oped into a spiral, while the other has elongated and broken up into very regular layers are labile. If a slight sinoidal vortices. On the convex side, near the right end, the beginning of the separation can disturbance is present, motions develop as shown in Figures 7-8. It is clearly seen be seen. No. 6 shows the permanent condition in which the flow begins to separate how separate vortices are developed by the mutual interference of the flows. The about at the narrowest cross section. vortex layer is rolled up inside these vortices, as shown in Figure 9. The lines of this Nos. 7-10 show the flow around a cylindrical obstacle. No. 7 shows the begin- figure are not streamlines, but such as were obtained by using a colored liquid. ning of the separation; Nos. 8-9, subsequent stages. Between the two vortices there I will now briefly describe experiments which I undertook for comparison with is a line of water which belonged to the transition layer before the beginning of the the theory. The experimental apparatus (Fig. 10) consists of a tank 1.5m (nearly separation. No. 10 shows the permanent condition. The wake of turbulent water 5 feet) long with an intermediate bottom. The water is set in motion by a paddle behind the cylinder swings back and forth, whence the momentary unsymmetrical wheel and, after passing through the deflecting apparatus a and four sieves b, enters appearance. The cylinder has a slot along one of its generatrices. If this is placed 78 Chapter 3: The Design Revolution

as shown in Nos. 11-12 and water is drawn out through a tube, the transition layer on one side can be intercepted. When this is missing, its effect, the separation, is eliminated. In No. 11, which corresponds, in point of time, to No. 9, there is seen only one vortex and the line. In No. 12 (permanent condition), the flow closely follows the surface of the cylinder till it reaches the slot, although only very little water enters the cylinder. A turbulent layer has developed instead on the flat wall of the tank (its first indication having appeared in No. 11). Since the velocity must diminish in the widening cross section and the pressure consequently increases (½ ρ v2 + V + p = constant on every streamline), we have the conditions for the separation of the flow from the wall, so that even this striking phenomenon is explained by the theory presented.

Translation by Dwight M. Miner, National Advisory Com- mittee for Aeronautics. Document 3-3(a-b) 79

Document 3-3(a-b)

(a) Alexander Klemin, Chapter X: “Resistance of Various Airplane Parts,” in Aeronautical Engineering and Airplane Design (New York: Gardner-Moffat, 1918), pp. 61-4.

(b) Grover C. Loening, Military Aeroplanes: An Explanatory Consideration of their Characteristics, Performances, Construction, Maintenance, and Operation (Boston: W.S. Best, 1918).

This pair of documents reproduces excerpts from two of the earliest aeronautical textbooks to appear in the United States, both in 1918. The author of the first text, Alexander Klemin, was a graduate of Massachusetts Institute of Technology’s (MIT) aeronautical engineering program (he actually succeeded Jerome Hunsaker as the instructor in 1913) and the head of the U.S. Army Air Service’s Aeronautical Research Department at McCook Field in Dayton, Ohio. The author of the second, Grover Loening (1888–1976), was an early aeronautical engineer and aviation manufacturer in the United States. He received an M.A. in aeronautics from Columbia in 1910, worked for Orville Wright, and served as chief aeronautical engineer of the Aviation Section of the Army Signal Corps at San Diego in 1914. Loening’s book first appeared in 1915, but it was not fully recognized until three years later when he had the benefit of evaluating the rapid pace of technological development that took place during World War I. In 1917, he formed his own company, the Loening Aeronautical Engineering Corporation, and designed a strut-braced monoplane for the army and experimented with amphibious aircraft. Loening won the 1921 Col- lier for his Air Yacht, a five-seat monoplane boat, which evolved into the Loening Amphibian, said to be the first practical airplane of that type. As might be expected, Klemin’s text was the more theoretical of the two, indica- tive of his advanced education in aeronautics at MIT. Loening, on the other hand, wanted a book that would more practically benefit “aviators and students” and that featured the lessons and experiences of military aircraft design, manufacturing, and operation. In important ways, the two texts represent what in the history of Ameri- can engineering education would come to be known as the “school culture” (Klemin) versus the “shop culture” (Loening). Shop culture, generally speaking, promoted a technical approach that could be applied directly to industrial work. In contrast, school culture placed more emphasis on basic studies in mathematics and science and on original research. To this day, some degree or another of tension between shop culture and school culture exists within most engineering communities and 80 Chapter 3: The Design Revolution Document 3-3(a-b) 81 has shaped the personality of American engineering schools and professional orga- Tractor Bodies nizations. Both approaches bear their type of fruit, although without question, the In Table 1 is given a comparative table of resistance coefficients for area in school culture approach has provided a much more solid basis for long-term funda- normal presentation of a number of airplane bodies, and in Fig. 1 are shown sketches mental development in aeronautics and most other highly technical fields. of the same bodies. Exact comparisons are impossible because some of the bodies One aspect that the two textbooks had in common was that they both reflected are made for two men and others for one. Still quantitative conclusions can be a state-of-the-art that had been largely driven by European developments in the drawn. The N. P. L. Model 5, more symmetrical than the B. E. 8, shows a distinct years leading up to 1918. Much of the information provided in the texts was based improvement over the latter which is somewhat discounted by the fact that the B. on knowledge and experience gained during the war in Great Britain at its National E. 8 carries two men unshielded. The B. Physical Laboratory in the 1910s, in France at Gustav Eiffel’s laboratory, or gleaned F. 36, all almost perfect dirigible form, from Germany’s many aeronautical advances. is markedly better than either of these two bodies. The resistance of the body in an air- Document 3-3(a), Alexander Klemin, Chapter X: “Resistance of Various plane is apparently a small quantity, but Airplane Parts”. the figures given below do not represent the resistance of a body in full flight Chapter X where it is increased by 40 percent, the propeller slip stream increasing the relative speed of the air by some 25 percent. Resistance of Various Airplane Parts Also, it must be remembered that with One of the most difficult problems in aeronautical design is the prediction of a best glide of 1 in 8, a 5-pound increase the total resistance of the machine. The wind tunnel test is a good check, but it is in resistance is practically equivalent to most important to assign resistance values to various parts and to tabulate them an added weight of 40 pounds. A blunt, prior even to the construction of the model. In this chapter have been collected as square form of body such as is often far as possible all the data available for bodies, radiators, fittings, wheels, cables and seen in American practice may increase wires and certain other miscellaneous objects. resistance even more, and better aerodynamical design of bodies seems a feature Airplane Bodies from the Aerodynamical Point of View worth considering. . If airplane bodies were designed from a purely aerodynamical point of view, they would follow dirigible practice and be of streamline form. There are, however, COMPARATIVE TABLE FOR TRACTOR BODIES a number of structural requirements which have to be met, which preclude the Pusher Bodies employment of such forms. The body must enclose the power plant and the person- A pusher body such as the Farman 3, illustrated in Fig. 2, gives a not much nel, the length must be long enough to place the rudders well clear of the wash of larger resistance than the tractor bodies, but when the head resistance of the uncov- the planes, the shape of the body must conform to structural requirements such as ered outriggers is taken into account, it will probably be found that, pusher arrange- the use of four longitudinal girders, or a triangular form which has been found to be ments offer considerably more resistance than tractor bodies. advantageous in steel construction. No wind tunnel tests on bodies alone can determine exactly their resistance on Radiator Resistance an airplane, because the question is complicated by the position and form of the The only values available for this are the results of some tests at the Massachu- motor and the disposition of the tail surfaces. The propeller in a tractor machine setts Institute of Technology. These were carried out on portions of a radiator of also introduces three possible variations in drag coefficients: (1) when the propeller is the honeycomb type having sixteen 1/4-inch cells to each square inch of the surface pulling and there is a slip stream of velocity greater than the airplane velocity, (2) the normal to the wind. The tests were repeated on two sizes of radiator section, one resistance on a glide when the engine is shut down, but the propeller is revolving as 0.25 square feet and the other 0.111 square feet, and at various speeds. No impor- an air motor, (3) when the propeller is not, revolving at all, the engine being held. tant variation in the resistance coefficient was apparent and the average coefficient 82 Chapter 3: The Design Revolution Document 3-3(a-b) 83

the French experiments was the fact that an uncovered wheel had a resistance of 50 percent more than a covered wheel of similar dimensions. This justifies the standard practice of covering the wheel in.

may be used for practical calculations. This has a value Kx = .000814 pounds per square foot of projected area per foot per second or .00173 pounds per square foot of projected area per mile per hour.

Resistance of Fittings Fittings are so variable in design that it is impossible to give definite figures to meet every type of wing strut fitting. Tests were conducted at the Massachusetts Institute of Technology on the fittings of which dimension drawings are given in Fig. 3; the coefficients of resistance are R = .00030 V2 and R = .00040 V2 for the two types which at 60 miles an hour gives 1.07 and 1.44 respectively. Such figures will be at least approximately correct in design.

Resistance of Airplane Wheels Resistance of Wires and Methods of Plotting For a standard airplane wheel of about 26 X 4 inches in size, the drag found by the N. P. L. is A certain complication is necessary in the methods of plotting the results for about 1.7 pounds at 60 miles per hour. This again the resistance of cables and wires. As we have seen from the diagram of Fig. 19, in is sufficiently accurate for practical purposes. Eiffel Chapter 3, of the Course, the resistance of a wire or any cylindrical body is partly has experimented with a number of wheels and due to turbulence, partly due to skin friction. It cannot therefore be represented by has shown that no great variation need be expected such a simple expression as front the above value. An important result from 84 Chapter 3: The Design Revolution Document 3-3(a-b) 85

R = KLD V2 F (VD). Resistance of Vibrating Wires When the question of resistance first began to arouse interest, it was popularly supposed that a vibrating wire had much greater resistance than a stationary one. We do not know what function F (VD) is exactly, nor how it varies with size This, however, is not the case. Research on this point at the N.P.L. failed to disclose and scale except from experimental results, and comparisons of resistance varying any difference whatever, although the balance would have shown deviations as small as LD V2 can only be made between two cables if VD is a constant. If K is taken as as 3 percent, even for the extremely small forces under consideration. Mr. Thurston, a function of VD, the R may be written R = KLD V 2 but then K must be plotted on the other hand, concluded that vibration at the rate of 15 per second increased against VD in analyzing experimental results. This is the only rational and scientific the resistance by about 5 percent for small wires and by a somewhat smaller percent- method. age for those of larger diameter. In any case, the effect is unimportant. An empirical method, however, is sometimes employed with fair accuracy of plotting the resistance of a wire whose length is equal to its diameter against V 2D2. Resistance of Stranded Wires This has the advantage that the graph approximates very closely to a straight line, The air resistance of stranded wires was also investigated at the N.P.L., and was the slope of which is equal to K, thus giving an easy means of determining a mean found to be about 20 percent greater than that for a smooth wire of the same diam- value of K. eter. This is only approximate, as the coefficient depends on the number of strands, type of lay, etc. It is also impossible to plot the values of K against VD for wire rope, Resistance of Stationary Smooth Wires as the VD law holds good only for objects which possess strict geometrical similarity,

The most accurate researches have been carried out at the N. P. L.R and their a thing which stranded wires of different sizes never do. K = results are, shown in Fig. 4 plotted against VD. In the expression LDV 2 , R is in pounds, L in feet, D in feet, and V in miles per hour. But in the abscissae, values of Resistance of Wires Placed Behind One VD, V’ is in feet per second, and D is feet, so as to give the correct scale and speed Another relationships which must be in the same units. The manner in which The accuracy of the curve at its lowest portion is doubtful, since the flow is resistance is affected by apparently just changing its nature at that point, and {AQ3} successive observations the close juxtaposition of under the same conditions may give quite different results. On modern machines two wires, one behind the of fairly high speed, however, the valves of VD nearly always exceed 0.35 and con- other, is a point of great sequently do not lie interest. Here, too, it is at on this section of the present necessary to rely curve. on Mr. Thurston, although Similar tests were we hope to be able soon to made by Mr. Thurston present the results of some and M. Eiffel, and the more extensive and accu- values obtained by the rate tests on this matter. former are plotted in the Fig. 5 gives, in terms same figure.Thurston’s of the resistance of a single experiments, however, bar, the resistance of two were very much earlier, bars or wires separated by and Eiffel’s covered a various distances. It will be less range and were per- seen that two wires placed formed with less sensi- one behind the other and tive apparatus, so it is spaced from 5 to 9 diam- advisable to use the N. eters apart, as is usual in P. L. results. double-wiring a biplane 86 Chapter 3: The Design Revolution Document 3-3(a-b) 87

cellule, have from 60 percent AIRPLANE WHEELS to 75 percent more resistance British Report 1912-1913, page 122. than a single wire. The force “La Resistance de l’Air et l’Aviation,” Eiffel, page 250. is, however, materially less WIRE AND CABLES than for the two wires placed “Aerodynamic Resistance of Struts, Bars and Wires,” by A. P. Thurston, Aeronautical side by side. Journal, April and July, 1912. Eiffel has experimented British Report 1910-1911. on the resistance of inclined “La Resistance de l’Air et l’Aviation,” Eiffel, page 97. wires. As would be expected “New Mechanical Engineers’ Handbook,” Section on Aeronautics, by J. C. Hunsaker. the resistance of a wire progressively decreases as its angle with the wind dimin- Document 3-3(b), Grover C. Loening, Military Aeroplanes: ishes. Table 3 gives correcting An Explanatory Consideration of their Characteristics, Performances, values. This table has been Construction, Maintenance, and Operation (Boston: W.S. Best, 1918). omitted by the editor.

Suggestions for Characteristics of Airflow. Stream-lining Having defined air, the manner in which it flows may be considered. Air either Wires flows smoothly past an object in streamlines—continuous filaments—or it breaks It has been suggested up into swirls and eddies, due to too abrupt a change in flow. The accompanying from time to time that photographs of airflow illustrate this. wire resistance should be decreased by “stream-lining” It is apparent that a spindle or fusiform shape, gently dividing the air at the or adding a triangular portion in back of the wire. From experiments by Ogilivie, front, and gradually permitting the filaments to close together at the rear, will give however, it appears that a section made up of a semi-circle a triangle has a decidedly a smooth flow, which amounts to the same thing as a very low resistance. It is also high resistance, and the gain from such a procedure wound be small. evident that a flat surface creates very great disturbance, and consequently high Wires placed behind one another have also been covered in. The British Royal resistance. Aircraft Factory produces a very heavy R.A.F. wire in use on big machines which is The curve of the streamlines, necessary to prevent disrupting them, may be stream-line in form. But the direction in which progress manifests itself at present computed for any speed, by applying fluid dynamics. But it must be kept in mind is in the elimination of wires by certain modern trussing such as used in the recent that a form of this kind gives its low resistance, only at one particular speed, since Curtiss biplane. the path of flow is affected by the speed. It is unnecessary here to take up the determinations of these forms. If the streamlines flow smoothly past an object, and close Resistance of Miscellaneous Objects up again without eddies, it follows that the only resistance experienced is frictional. This resistance of certain miscellaneous objects as deduced by Eiffel may some- There is hardly any shape, however, which does not create small eddy resistance. times be useful. The values for such objects within certain limits are illustrated in Methods of measuring the resistance of the air that have been widely used, are Fig. 6. the following: 1. Dropping surfaces from a height and measuring time of drop and pressure, References for Part I, Chapter 10 used by Newton, and Eiffel in his earliest experiments. AIRPLANE BODIES British Report 1911-1912, page 52. 2. The whirling arm, used by Langley, and consisting of whirling the surface at British Report 1912-1913, page 116. the end of a large arm around a circle of large diameter and recording the resistance “La Resistance de l’Air et l’Aviation,” Eiffel, 1914, page 250. automatically. 88 Chapter 3: The Design Revolution Document 3-3(a-b) 89

3. The moving carriage, an automobile, trolley or car, as used in the experi- ducing smoke or particles into the air and photographing it. Ammonium Chloride ments of the Duc de Guiche, Canovetti, and the Zossen Electric Railway tests. is a very convenient smoke.

4. By blowing or drawing air through a tunnel in which the object or a model of Importance of Visualizing the Air. the object is placed. This method is the most modern and convenient, and permits of It is of great value in aeroplane work, to become accustomed to visualize the a uniformity of the air current, which cannot be obtained as easily in the open. streamline flow of air, and ability to “see the air” often solves many problems of stability and reduction in resistance, without any recourse to mathematics or measure- In wind tunnels the best practice is to draw the air in, through screens and ments. Besides this, there is offered in the study of airflow by photography, a field channels that straighten it out, past the experimental chamber, and thence to the of investigation of great promise and absorbing interest. fan. Practically all the great Aerodynamic Laboratories use the wind tunnel method of experiment. The prominent ones are: the Eiffel laboratory in Paris, the National It is a common experience that in a wind, at the front of a flat surface, there is a Physical Laboratory in England, and the tunnel at the Washington Navy Yard. The dead region of air, where no wind is felt. Photographs show this air cushion clearly. speed of the wind in the In stability dis- Eiffel laboratory can be cussions, effect of fol- brought up to almost 90 lowing planes, inter- miles per hour (40 meters ference and propeller per second), and its size stream action, price- permits of testing many less secrets would objects, such as struts, be revealed if the air to full size and complete could be followed in models of aeroplanes to its every movement. one-tenth full size. Such a magnitude permits of exceedingly valuable determinations, and the work of Determination of Air Resistance. the laboratories is daily being applied with entire success to full-sized aeroplanes, The nature of the action of air on objects has been considered, but we must although the higher speeds of aeroplanes require considerable correction of wind know in addition with what force in pounds P, the air pushes on an object when it tunnel results. This is particularly true in the measurement of pressures on wing passes it at velocity V. We cannot refer to theory for this, satisfactorily, so we must sections at low angles. obtain actual measurements of the air resistances on various objects. It must be borne in mind, therefore, that the air in a tunnel is confined and that 2. Air Resistance Increases as the Object’s Size Increases. all tunnel results are not perfectly adaptable to machines unless suitable corrections are applied. Combining the results of all the laboratories, we may draw some general con- This experimental fact is also subject to modification, since, as the size of sur- clusions with regard to air resistance, as follows: face increases, the pressures are somewhat greater in proportion. But we can disre- gard this also without serious error. 1. The resistance of an object in an airstream is proportional to the square of the velocity of the air. 3. The Air Density Influences the Air Resistance.

In other words, if the velocity is doubled, it follows that the resistance will be It has already been pointed out that heavy air (low altitude) has more resistance increased four times, or if velocity is five times as great, the force on the same object than light air (high altitude). would be twenty-five times as great. This is merely an experimental fact. 4. The Shape of an Object Controls its Air Resistance. There are many ways of determining the manner in which the air flows past an object, such as noting the direction in which light silk threads are blown, or intro- The beautiful streamline photographs have already discovered this for us, and 90 Chapter 3: The Design Revolution Document 3-3(a-b) 91

show how easily and with what small resistance the air slides by a streamline shape.

Let us combine all this into a compact sentence called a formula, where P = the resistance in lbs., S = the area in sq. ft., V = the velocity through the air in miles per hour, d = the density of the air in lbs. per cubic foot, and finally describe the shape of the object in order to ascertain whether it is clumsy or streamline, by a numerical multiplier, which we will call k, and which we will define as a “shape coefficient.”

In others words, P = k. d. S V2.

But to simplify matters, since all of these shape coefficients for various shapes have had to be measured, and mostly at sea level, we can call d also a numerical factor, and combine it with k, so that kd=K, which is a number, always applicable to that particular shape, and represents the coefficient at sea level, in which we can use, for any size body of that shape, at any speed, in order to obtain the resistance at sea level. For many different shapes, all we need, therefore, is for someone to measure and tell us what the values of K are for different shapes. Just as a grocer will tell you that a piece of cheese weighs half a pound because he measures it, so has M. Eiffel told the aviation world that K for a wire is .0026. Fineness Ratio — is a term used to define the general shape of bodies, and is 2 Therefore P = KSV obtained by dividing the fore and aft length of the body by the greatest width across the wind. This formula applies to all air forces, whether they are resistances, as we are considering here, or lifting pressures as we will consider later—with this reservation, Master Diameter — is the greatest width of a body across the wind. that the shape coefficient is either for resistance, or for lifting power, and one must always be careful to determine which it was meant to be measured for. Fairing — is used to denote the additional “tail” or filler used to make a poorly We may now proceed with the interesting study of what values of K have been shaped body more streamline in form, thereby reducing its resistance. found for various shapes, and not only will those be described by diagrams, but fre- Diametral plane —is the plane, passed through a body, facing the wind perpendicu- quent numerical examples are given, of how to apply the data to determine values larly, and cutting through at the master-diameter. of air resistances. A study of this subject will give the student a most valuable insight into this branch of the science, and an appreciation of what shapes are “good” and Normal plane — is another expression for diametral plane, and merely refers to the “bad.” maximum cross-sectional projection of the body. It also refers to a flat surface held normal (perpendicular) to the air current. Equivalent Normal Plane — is the size of normal flat surface, that would give the Definitions. same resistance as does the body referred to. In Aerodynamical studies it has become customary in defining objects to use unfa- Flat Surfaces, miliar terms. Normal to the Airstream. Aspect Ratio — is a term used to define the shape of a surface, and is the long span Square Planes: of the surface across the wind divided by the width. In square planes, normal to the air, the value of K is .003 for surfaces up to two or three feet square, and .0033 for very large surfaces like the sides of buildings. 92 Chapter 3: The Design Revolution Document 3-3(a-b) 93

It may be stated, therefore, for aeroplane usage, that P, the air resistance in lbs., Discs: of a square surface, S sq. ft., in area, at a velocity V miles per hour, is The shape of flat surfaces also affects their air resistance. Passing from a square plane to a round disc, reduces K to .0028, so that the air resistance of a disc 2 feet P = .003 S V2 in diameter, at 60 miles per hour, is

Thus, for a surface 2 feet square, at 70 miles an hour: P = KSV2 = .0028 X .7854 X 4 X 3600 P = 32 pounds P = .003 X 4 X 4900 P = 58.8 pounds In general rounded edges may be expected to reduce K, for flat surfaces. Discs or flat rectangles, placed one in front of the other, interfere with each Rectangles: other and exhibit a most important phenomenon, shown on page 52. This aspect ratio of a square is one. Rectangles have aspect ratios above one, Cylinders: when presented normally to the air. Passing from the disc to the cylinder, with the circular base facing the wind, the Up to an aspect of 5 or 6, K remains about .003. resistance is found to be less as the length of cylinder is increased, until the length An increase in the value of K is found for rectangles as the aspect ratio increases. becomes greater than 5 diameters, when the resistance is found to increase again. When the aspect ratio of the rectangles increases to 15, K becomes .0035 and Some values of K are given on the chart on page 48. on further increasing the aspect ratio to 30, K = .0038. Wires and cables are merely long cylinders. Extensive experiments have been A flat rectangle, perpendicular to the air current, with its dimension across the conducted on them, and values of K found. For smooth wires K = .0026, whereas current, thirty times as large as its width, might be met with in rods, temporary cables are found to have considerably higher resistance with K = .003. struts, etc., and it is interesting to note how high the resistance would be. Thus, a machine having 200 feet of 1/8 inch cable, giving a projected area of 200/96 = 2.08 sq. ft., will have an air resistance due to the cables at 80 miles an hour of

P = .003 X 2.08 X 6400 = 40 lbs.

This high value immediately suggests the advisability of reduction of cable resistances. In double cables, it would prove beneficial to tape them together, so as to streamline each other. A graph is given showing the reduction in resistance due to inclining the wires i.e., staggered planes, on page 48.

Spheres: The resistance of the air on spheres presents a study of interest. The sphere is the simplest geometrical form, and, as a basic one, it should long ago have served as the unit form for air resistance. Lack of agreement in the experimental results of different laboratories was only cleared up when Eiffel discovered that an increase of speed of the air above 20 miles per hour caused a change of flow, due to the flatten- ing out of vortices back of the sphere, which reduced the resistance considerably. And that above this speed, the nature of the air resistance remained constant. K = .00044, for a sphere, at speeds above 20 miles per hour, whereas at very low speeds K becomes .001. In having a smoother flow at the higher speeds, less lbs. of air are put in motion, which means that the resistance is less. This action of air, in tending to smoother flow with speed increase, is important to bear in mind. 94 Chapter 3: The Design Revolution Document 3-3(a-b) 95

Streamline Shapes: In this class may be included bodies of fusi-form or streamline form, shaped for least resistance. Their application to the design of tanks, fuselages, nacelles, hoods, etc., is of fundamental importance. In a most interesting set of experiments, conducted by M. Eiffel, on streamline shapes, illustrated in the diagrams and chart on p. 52, the bodies consist of a nose, a cylindrical central portion, and a tail.

The results of the experiments show that:

1. The blunter the nose, the greater the resistance. 2. The shorter the central cylindrical portion is, for the same nose and tail, the lower the resistance. 3. The effect of shortening up the tail is not very great, although slightly increasing the resistance.

In each case, however, measurements made at speeds up to 90 miles an hour

for dirigible balloons which it is important to consider. The models tested measured 3.75 feet long and .62 feet in diameter, giving a fineness ratio of 6. The shapes in their order of least resistance and values of K for 25 m.p.h. are given. At higher speeds, still lower Ks would be expected. showed that the resistance does not vary as V2, the value of K becoming constantly less with speed increase. This is a very significant determination, and may be The form No. 1, having the least resistance, is, perhaps, the best form that has explained on the ground that, in bodies of this kind, the major part of the resistance ever been tested in a laboratory, and at high speeds would give a resistance about 2 at high speeds is frictional and therefore increases at much less than V . In addition 1/25th of the normal pressure on its diametral plane. It is the form used in the Par- the effect of velocity increase is to flatten out the flow and suppress eddies. seval non-rigid dirigibles.

The values of K for these bodies are given. It is interesting to note in studying low resistance bodies, how closely they resemble the shapes of fishes, and of birds, measurements of a fast swimming fish The Goettingen Laboratory conducted extensive experiments on the best shapes showing an almost exact resemblance to this Parseval shape. 96 Chapter 3: The Design Revolution Document 3-3(a-b) 97

As a general rule, the best streamline body is the one having a fineness ratio of 6 and with the master diameter about 40% back of the nose, both nose and tail being fairly well pointed.

Struts: The application of fineness ratios, and shapes of least resistance, to improvement in the form of struts, has in many instances tremendously improved the performance of aeroplanes.

In addition to the form for least resistance, however, the weight of the struts and their strength are factors that must be considered in choosing the best shapes. We will confine ourselves here, however, to a study of the resistance of various shapes.

A group of strut sections are given and K for each one. It is to be noted that the effect of yawing is greatly to increase these resistances by presenting the strut sidewise to the air, and it will be necessary later to consider the amount of this increase.

Inclining the strut to the vertical, as in staggered planes, has the effect of increasing the length of section in the airstream, and, consequently, the resistance does not decrease for streamline shapes, while for blunter shapes, inclination reduces the resistance considerably.

In struts, as in bodies, an increase of velocity is accomplished by a reduction in the value of K, that is more noticeable the greater the fineness ratio, i.e., the longer the section of the strut. This is again due, probably, to the preponderance of friction in the total resistance.

The results obtained in studies of strut resistance indicate the importance of having struts well made and of a uniform section. Just as in bodies, abrupt changes in contour must be avoided and attention paid to a smooth curve on either side of the central portion. It is found, in general, that a fineness ratio of 5 to 1 is best for use, where a fin effect is desired, and where not,— the best fineness ratio is 3 to 1. Wheels: The air resistance of chassis wheels is a considerable item in flight. Experiments have been conducted on various-sized wheels, and the results are as follows:

28 ½ inches diameter by 2 ½ inches tire, K = .0025 24 ½ inches diameter by 3 ½ inches tire, K = .00265 21 ½ inches diameter by 3 ½ inches tire, K = .0018 18 ½ inches diameter by 2 ½ inches tire, K = .0021 98 Chapter 3: The Design Revolution Document 3-3(a-b) 99

When the wheels are covered in, it is found in almost every case that the resistance is halved, so that for the 24 inch X 13 inch wheel, when covered in, K = .00133. An average K for wheels would be .002.

As an example, it is desired to determine the resistance of two 26 inch X 4 inch wheels at 80 m.p.h.

The projected surface = 1.4 sq. ft. ∴P = .002 X 1.4 X 6400 = 18 lbs.

If the wheels were covered in at this high speed, about 9 lbs. would be saved in resistance; this would permit of carrying about 60 lbs. more load on an efficient machine, or would add 10 gallons more fuel.

Summary: The data given in this chapter enables the air resistance of various shaped bodies to be computed for any speed V and any size surface S, where S in the maximum cross-sectional projection of the body, perpendicular to the airstream. It is merely necessary to supply the numerical values of K, S (in sq. ft.), and V (in m.p.h.), in the formula.

P = KSV2

Where K = kd, d being the density of the air, and the values here being correct only for sea level and therefore largely comparative. It is well again to recall that the propeller of an aeroplane must give a pull or push great enough to overcome:

1. The resistance to motion of the struts, wires, body, wheels, fittings, skids, gas tanks, etc., called Structural Resistance. 2. The dynamic resistance of the wings and rudders, called the Drift and generated by the same pressure that gives the Lift.

In this chapter the first has been considered. And a study of the second may now be taken up. 100 Chapter 3: The Design Revolution Document 3-4 101

Document 3-4

Louis Breguet, “Aerodynamical Efficiency and the Reduction of Air Transport Costs,” Aeronautical Journal (August 1922): 307-320.

In a speech delivered before the Royal Aeronautical Society in 1922, French aircraft builder Louis Breguet made one of the strongest cases for the social and economic benefits of aerodynamic streamlining. The British Aeronautical Journal subsequently published his talk, which is included in its entirety below. The theme of his talk, as indicated in its title, was how improved aerodynamic efficiency could reduce the cost of air transport and thereby breathe life into civil and commercial aviation. To achieve this goal, engineers would need first and foremost to improve an airplane’s “fineness ratio,” by which he meant its lift-to-drag ratio. Improving this ratio would extend the range of aircraft and make their overall operation more efficient. The way to achieve this improvement was through the use of retractable landing gear and other forms of aerodynamic streamlining. Breguet even set a target for the fineness ratio of commercial aircraft, one that would be hit barely a decade later by the Douglas DC-series aircraft. The aircraft that Breguet built after 1922 crept closer and closer to such levels of aerodynamic efficiency. The most remarkable of his designs, the Breguet XIX, which was refined from a 1921 military biplane, “enabled France’s globe-trotting pilots to start an international craze for nonstop long-distance flying” (Terry Gwynn-Jones, “Farther,” in Milestones of Aviation [Hugh Lauter Levin Associates, Inc., 1989], p. 46). Two of these flyers, Dieudonné Costes and his navigator Joseph Le Brix, in a Breguet XIX, made the first nonstop crossing of the south Atlantic in October 1927, five months after Lindbergh’s historic flight thousands of miles to the north. A year earlier, in 1926, Breguet XIXs broke the nonstop flight record three different times; in the last, pilot Peltier D’Oisy flew a Breguet nearly 6,300 miles from Paris to Peking in just a little over two and one-half days. 102 Chapter 3: The Design Revolution Document 3-4 103

Document 3-4, Louis Breguet, “Aerodynamical Efficiency and the Reduction of Air S is the projected area of the wings on a plane parallel to the direction of Transport Costs,” Aeronautical Journal (August 1922): motion. 307-320. V is the speed of the aeroplane along the trajectory.

R, = K,S V- (2). PROCEEDINGS. wherein :— TWELFTH MEETING, 57th SESSION. R, is the lift of the aeroplane, normal to the direction of motion. A meeting of the Society was held in the rooms of the Royal Society of Arts, K, is a- coefficient relating to R, and dependent upon the value of the angle of Adelphi, London, on Thursday, April 6th, 1922, the Chairman (Lieut.-Colonel M. incidence in flight. O’Gorman) in the chair. R= = (K= + (T/S) SV2 The CHAIRMAN said that the members were to hear M. Louis Breguet, whose name they were too well acquainted with to need any introduction from him. The Dividing equation No. (r) by equation No. (2), which gives the ratio of drag to members ought to know, however, that M. Breguet had been decorated in 1911 and lift, we have: subsequently promoted to the rank of officer of the Legion of Honour, in France, for the services he had rendered during the war, and moreover, that he was the Presi- R./R, = (Ks + a-IS)IK9 = tan (3) dent of what corresponded, in France, to the Society of British Aircraft Construc- tors, namely, the Chambre Syndicale des Industries Atsronautiques. (Applause.) That expression tan 0 is usually tailed the “fineness” of the aeroplane, and every The CHAIRMAN then called upon M. Breguet to read his paper on “Aero aeroplane is characterised by a certain minimum value of tan ¢ which can be repre- dynamical Efficiency and the Reduction of Air Transport Costs.” sented by tan 0. The smaller tan 0m is the better will be the aerodynamical qualities of the aeroplane. AERODYNAMICAL EFFICIENCY AND THE REDUCTION OF AIR That fineness is used when calculating the longest distance which a given aero- TRANSPORT COSTS. plane can fly in a “no wind” atmosphere. The possibility for aeroplanes used in air transport services to be made to pay has, so far, been very often questioned, considering the present cost price per ton- L = (622 pl m tan ¢m) log PI (P - p) (4) mile flown. I intend to demonstrate that one can, from now, predict an important reduc wherein :— tion in air freight rates, not only through greater safety of flight mainly resulting L is the said longest distance in kilometres. from a longer life of aeroplanes and engines, but also through the betterment of p is the efficiency of the propeller. several coefficients which characterise the aerodynamical qualities of aeroplanes. m is the “fuel and oil” consumption of the engine per horse-power and per hour The study of these coefficients can be made either by laboratory tests on models in kilogs. or by full-scale tests in flight which, if judiciously interpreted, should yield informa- P is the total weight in kilogs. of the aeroplane and cargoes at the start. p is the tion sufficiently accurate for practical purposes. weight in kilogs. of the “fuel and oil” consumed by the engine during the flight and May I remind you that the principles of aerodynamics lie in the following for- which was included in P. mulae:— One at once realises the very great importance of the fineness tan ¢m which in that formula is the only term depending upon the aerodynamical qualities of the Rx = (Kx + o/S) SV2 ….. (I){ aeroplane. For certain given values of L, p, -in and P, a reduction of tan ¢m causes a reduc- Wherein:— tion in the “fuel and oil” consumption, and consequently increases the weight of Rz is the total drag of the whole aeroplane along the direction of motion. paying freight which could be carried. Kz is a coefficient relating to R= and dependent upon the value of the angle of The conclusion is that one must bring to the minimum the value of tan ¢m. It incidence in flight. can be obtained by choosing the best possible profile for the wings, the best designs a- is the projected area, normally to the direction of motion, representing the for the body, empennage, etc. Moreover, the undercarriage should be made to disap- passive resistances of the aeroplane. pear inside the body or the wings when the aeroplane is in flight, etc. 104 Chapter 3: The Design Revolution Document 3-4 105

Another interesting coefficient appears when one calculates the power spent in than one half the present rates of aerial freight, and how also, when bringing other horizontal flight is given by the formula :— improvements which I will specify later on, one can expect to bring down the rates of aerial freight to the value of those now charged in France for first class railway 3 W = RX V = (K + σ/S) SV passengers. An aeroplane of high standard quality now has :— Where A fineness equal to ...... 0.12 Its coefficient of minimum power is equal to...... 0.55 W is the power in kilogrammetres Its propeller efficiency is ...... 0.73

RX is the drag in kilogs as calculated by formula (I) S is the wing area in square metres Fuel and oil consumption of its engine at an altitude of 2,000 metres is 290 V is the speed in metres per second grammes per horse-power and per hour. We will, moreover, admit the total weight of that aeroplane to be such as to allow it to climb to 4,500 metres within one and a half hours if the engines are On another side, the value of the lift, as given by formula (2), is equal to the run at full power,* and we will consider an aerial line of 800 kilometres non-stop total weight P of the aeroplane in horizontal flight. We thus have : flights. If such an aeroplane has a wing surface of 100 square metres, for instance, and 2 P = Ry = KySV a power plant of 600 horse-power, its dead weight will be equal to 2,200 kilogs., and it will be able to carry a total load of 2,100 kilogs. Its total weight will thus be Eliminating the speed V between the two equations (5) and (6) we come to equal to 4,300 kilogs. another formula giving the power W: Its flying speed at full power at an altitude of 2,000 metres will reach 175 kilometres an hour, and its commercial speed at the same altitude can be reckoned as to 3/2 W = P √(P/S) (KX + σ/S/(Ky) be equal to 150 or 155 kilometres an hour. The weight of fuel and oil necessary in a still atmosphere for a given flight is We thus see that the necessary power is in direct proportion to the value of the term deduced from the formula (4), which, in the present case, gives the weight as being 568 kilogs. 3/2 (KX + σ/S)/(Ky) But practical flying has proved that one must expect to have to fly against a fifty kilometres an hour wind, and therefore it is necessary to have at least a fifty percent. I have represented the term by the Greek letter S and call it the coefficient of safety margin in fuel and oil. In other words, we shall have to carry in the case under power. consideration a total weight of fuel and oil equal to 568 x 1.5 = 850 kilogs, whilst For a given aeroplane, the value of % will reach a minimum for a certain angle the average consumption will only be of about 1.1 times the calculated quantity of of incidence in flight, and to that particular value of S will correspond the minimum 568 kilogs., that is to say, 625 kilogs.f of power required, and that is why I propose to call that special value S,n the coef- The crew (pilot and second pilot-mechanic) together represent a weight of ficient of minimum power. about 180 kilogs., and another 70 kilogs. are necessary for various instruments and That coefficient of minimum power is used to calculate the minimum power T.S.F. apparata. necessary to maintain in horizontal flight a given aeroplane, and the height of ceil- The total weight of fuel and oil, crew and instruments thus reaches eleven ing chosen will fix the nominal power to be provided for the engine. One can hundred kilogs., leaving a clear margin of one thousand kilogs. for passengers and therefore see that, should it be possible to reduce in the proportion of two to one, paying cargo. for instance, the value of Sm, then an engine of half the nominal power would be The present running costs of an aerial line of 800 kilometres long, with the sufficient to reach the same ceiling. above aeroplanes, are as follows :— Since I have now clearly shown the way these several coefficients affect the aero- Francs. dynamical qualities of aeroplanes, I can explain to you how one can design in the Petrol (820 litres at francs 1.82 for Boo kilom. flown) per kilom. 1.87 future, starting from existing aeroplanes, new ones which should reduce by more Oil (50 litres at francs 3 for Boo kilom. flown) o. ig 106 Chapter 3: The Design Revolution Document 3-4 107

Crew (pilot at francs 0.20 per kilom., second pilot-mechanic It is evident that the reduction in power is only due to the betterment of the at francs 0.15)* 0.35 coefficient of minimum power and of the efficiency of the propeller. • Sinking fund for aeroplane and engine ...... 6.50 One can thus spare 312 horse-power, which means a saving of 468 kilogs. of Upkeep of aeroplane and engine and aeroport expenses, etc. ... 4.50 dead weight (at the rate of one and a half kilogs. per horse-power for the engine and General expenses of the company 4..00 its appliances). Francs 17.41 The quantity of fuel and oil to be carried can be deduced from the formula (4), Should an aeroplane always fly with full cargo load and without incident or as has already been done for the first type of aeroplane considered. This gives interruption, the cost price per ton and per kilometre would thus be about 17.40 p = 230 kilogs. francs. But one has to reckon with trips made with smaller paying loads, or inci- The weight it is wise to carry up will be taken equal to dentally interrupted, and therefore it is wise to calculate the cost price per ton on 230 x 1.5 = 345 kilogs. flights with half cargo loads as an average, which brings the present cost price per and the average actual consumption will be about ton and per kilometre to 35 francs. That is about the price at which are now work- 230 x 1.1 = 255 kilogs. ing the best aerial lines. A few companies work on still higher prices, either because The weight saved on fuel and oil with the second type of aeroplane will thus be the average cargo load is less than one half the maximum, or because they fly over practically particularly awkward grounds. They then reach such prices as 50 francs or £I 12.s 850 - 345 = 505 kilogs. od. per ton-mile. and the excess of useful load is thus increased to a total of Should such prices be considered as irreducible they would be practically pro- 468 + 505 = 973 kilogs. hibitive and little chance would be left to commercial aviation, since one would The commercial efficiency of that type of aeroplane will thus be practically have to charge from eleven to twelve hundred francs per passenger from Paris to double that of the first type, since it will be able to carry 1,973 kilogs. of passengers London or vice versa, without any profit to the company. or paying freight, instead of 1,000 kilogs. How can these figures be reduced in the near future? We will consider:— At the same time that aeroplane will cost about 25 percent less than the first, 1. —Advantages to be drawn solely from the betterment of aerodynamical qual- since its nominal power plant will only be half the size of the first, and the value of ities of the aeroplanes and of the thermal efficiency of engines. the nominal power plant now represents about half the price of an aeroplane. 2.-Advantages resulting from such improvements in the engines and in mechan- The running cost of the new type of machine will then be as follows:— ical parts of aeroplanes as would bring reductions in the provision for sinking funds, Francs. upkeep of machines and general expenses. Petrol (335 litres at francs 1.82 for 8oo kilometres) per kilom. 0.765 1.-Advantages to be drawn solely from the betterment of aerodynamical quali- Oil (20 litres at francs 3 for 8oo kilometres) 0.075 Crew ties of the aeroplanes and of the thermal efficiency of engines...... 0.35 Sinking fund (75 per cent. of francs 6.50, since the cost Taking, for the sake of demonstration, the same size of aeroplane as already price considered, that is to say of the aeroplane has been found reduced in that proportion) ...... Wing area: 100 square metres. 4.90 Upkeep of aeroplane and engine and aeroport expenses 4.50 Gen- Total weight: 4,300 kilogs. at the start. eral expenses of the company 4.00 R Ceiling: 4,500 metres. Francs 14.59 And if we suppose that we can bring down As the paying load has been raised to 1,973 kilogs., the cost price per ton and Its fineness to o.o65 instead of 0.12 per kilometre becomes equal to 7.40 francs, or—if calculated on a half cargo load Its coefficient of minimum power to 0.28 0-55 basis—14.80 francs instead of the present cost price of 35 francs. London to Paris Its propeller efficiency to 0.775 0.73 The fuel and oil consump- passenger fares can then be brought down to some 450 or 500 francs without profit tion of its for the company. Although very high, these last figures are more encouraging and engines per horse-power and per nearly workable. hour to 215 ego grams 2.-Advantages resulting from such improvements in the engines and in mechan- then such an aeroplane will not require more than 288 horse-power to reach the ical parts of aeroplanes as could bring reduction in the provisions for sinking funds, same ceiling of 4,500 metres instead of 600 h.p. upkeeping and general expenses. 108 Chapter 3: The Design Revolution Document 3-4 109

When one can rely on an average life of one thousand hours for aeroplanes can build very good aeroplanes of four to five tons total weight; but I think it would and engines instead of the present 200 or 250, much better prices will be obtained. be much nearer to reality to talk of 500 square metres wing surface, twenty to forty Supposing that. such an aeroplane be bought for the same price as the last one tons total weight and fifteen hundred to four thousand horse-power. Such large mentioned, but calculating on 1,000 hours life, the sinking fund only calls for 1.10 machines* will most probably have very thick wings. By thick, I mean about six francs, whilst the upkeep can be reasonably considered as half as expensive since feet, and within these wings will be provided cabins, saloons and every comfort for the aeroplanes and engines will be of much better quality and strength. General passengers. It is worth noting that if a large increase of the size of an aeroplane does expenses will also be considerably reduced through the much larger turnover that not improve materially its commercial efficiency, it has nevertheless the great advan- the companies will then be able to secure. tage of allowing much more room for passengers and freight, and that can be easily In such conditions the running costs would be as follows: Petrol understood when one remembers that whilst the weight, the power, the capacity in per kilometre, francs 0.765 Oil ...... 0.075 Crew ...... cargo and the cost price of an aeroplane vary as the square of its lineal dimensions, ...... 0.35 the volume to be reserved for passengers will vary as the cube of the same dimen- Sinking fund r.r0 sions. In other words, if the weight, surface and power have been increased one hundred times, then the cubic capacity will be one thousand times larger.* Passengers, Upkeep ...... 2.25 for instance, will proportionately dispose of ten cubic metres each instead of one General expenses (estimated) r.oo cubic metre in the first case, and that shows one of the most important advantages Francs 5-54 to be drawn from the manufacture of very large machines. That is to say, 5.55 francs for 1,973 kilogs. carried, or 2.82 francs per ton and It is not exaggeration to talk of the days when the price of aerial transport will per kilometre, or 5.65 francs if calculated on the usual half cargo load basis. Then come down to two francs per ton and per kilometre, since an average life of two or the Paris to London fare will become about 190 francs, or say £4, the present price three years—or say two thousand hours of flight—for aeroplanes and engines would of the third class railway fare. suffice to bring that result, even with petrol and oil at their present very high prices. When that time comes then the aerial transport companies will be flourishing It is practically certain that aeroplanes of the future will not burn petrol but rattier paying concerns, and no more subsidies will be required from the State, because the heavy oils or other cheap fuels, the cost price of which should be about one-fifth time saved in travelling by air from Paris to London, for instance, will certainly be that of petrol. worth a fare of say £6, showing £2 profit on each passenger. Now first class passenger fares in France are calculated on the basis of 21.15 It is worth noting in the last cost price we have given that fuel and oil, although francs per one hundred kilometres. Supposing that a passenger with his hand lug- intrinsically very expensive, amount to only fifteen percent of the total cost price, gage weighs an average of ninety kilogs., we find that the ton-kilometre (passenger) whilst the crew draws 6.4 percent. runs to 2.35 francs. That shows that the future prices of aerial transports will be of On the other side, sinking fund, upkeep and general expenses, which, on the the same order as the present first class passenger fares in France. present cost of 35 francs, represent 86 percent of the total cost price, have grown More striking still is the comparison with the fares on steamship lines, since first to 92 percent with the second example and come down to 78 percent in the last class passenger nowadays pays about six francs per ton-kilometre, whilst state cabins considered circumstances. are charged at the rate of ten francs. May I therefore state—in opposition to the sayings and writings of certain In conclusion, aerial transports are very expensive for the present because they experts who do not know much about aerial transports—that it is not the cost of are not yet out of the experimental stage and that the sinking funds, the upkeep and fuel or crew which makes aeroplanes expensive machines, but only their present general expenses are very heavy; but one can reasonably say that within ten years short life, with resulting consequences. But we must expect and hope to see large these costs should be reduced in the proportion of seven to one, and within twenty and strong aeroplanes of the future become as safe as motor cars, steamships and years roughly in the proportion of fifteen to one. railways, and when that time comes, then we will be able to apply to sinking funds Moreover, one must not forget that “time is money,” and therefore a saving of the same coefficient as is now used for steamships, that is to say, about an average four or five days on London to Cairo, for instance, will be of tremendous value to business men. The fares charged will then be of comparatively small importance to life of twenty-five years! them as long as safety and speed are secured, and as aeroplanes will be unbeatable From the example I have taken for the purpose of demonstration of an aero- as regards speed, aerial transports are bound to wipe out all other systems of interplane of one hundred square metres wing area and 4,300 kilogs. total weight, you national communications. must not infer that I expect the improvements of aerodynamicaI qualities (with the We must therefore work hard and steadily, with full confidence in the future of consequences I drew from it) to be realised by such small machines. I believe one aviation. 110 Chapter 3: The Design Revolution Document 3-4 111

DISCUSSION. had at least produced a very lengthy correspondence from the inventors. In having The CHAIRMAN said they had listened with interest to M. Breguet’s display M. Breguet at the meeting the Society was greatly honoured. He (the speaker) was of an optimism which was well worthy of their respect. He had indicated the steps an optimist; his optimism was not based merely on the feeling that the success of by which they might compete on equal terms with railways for certain classes of air transport depended solely on the research laboratory, which the President had traffic, and if they did not arrive at the whole result they might look step by step at emphasised. It had been very much borne in upon him during the last few days that the various factors and see which of them they had failed to bring up to the high even with the latest and best of up-to date machines they did not necessarily achieve level which had been foreseen. He had suggested that the fineness of aircraft could an air service. An air service was dependent, quite apart from equipment, on the be improved by 50 percent. He had shown the hope that fuel consumption might ground organisation that dealt with it. If we were able to make machines that we be reduced, and had discussed similar factors, showing that even if these factors of could run at one-third, or, one-fifth, or even one-fifteenth of the present-day cost, improvement could not all be obtained in toto, he had shown that the sum of the that would not make possible a flight to Paris in thick fog and low clouds, and it advances obtainable make a fundamental and logical improvement in the whole would not enable the pilot to know whether he could start now or in an hour’s aeronautical situation. M. Breguet’s optimism was legitimate if the moneys spent time unless the meteorological service was very good. It was the improvement of on research were not reduced. Each specialist could then tackle, in his own sphere, the adjuncts to an air service, quite apart from the machine itself, or the skill of the some one of the problems that M. Breguet had exposed, and in such measure as mechanics who looked after it, or the pilots who flew it, that would make develop- they secured improvement, these would become operative together, and we should ment possible. “If it were possible to fly present-day machines in all weathers strictly approach the competition, on an equal basis of charge, of first class railway fare and to scheduled time, and for them to return strictly to scheduled time, no matter what travel by aeroplane, with the enormous advantage to the aeroplane traveller that the weather was, costs would be very considerably reduced as well as charges to the he had saved so much in time. With regard to the author’s prophecy of the 40-ton public. He had looked with great interest at the formulae in the paper and he did monoplane, he did not suppose that an engineer of the author’s distinction would not know whether there was a solution to any of those equations that gave one the have put that forward entirely at random without having thought it out as a thing value of profits. It seemed to him that if they could determine that accurately by thoroughly worth considering as a commercial possibility. The author had shown means of an equation it would be an excellent thing for mathematical treatment. that the real incoming of aerial transport was not going to be achieved by prophetic As a distinguished mathematician had said, “ What comes out of the mathematical letters to the Press nor yet by meetings of the Civil Aviation Advisory Board at the mill depends on what you put into it,” but there were so many variables in arriving Air Ministry, unless the latter proved themselves competent to understand the tech- at a result by mathematical methods that it was necessary to study the component nical nature of the problem and secured a sufficient addition of funds for the spe- factors entering into it before they could arrive at a figure for the profits they were cialised research which civil transport needed. He agreed that civil transport was the likely to get. He laid stress on the particular item of the ground. Organisation. He basis for full factories and full factories the basis of military aeronautical operations. was very much interested to see, in the latter part of the paper, M. Breguet’s visu- Nothing else but the scientific work of men of intelligence and intellect in the labo- alisation of the very large aeroplane. The difficulty of structure weight had always ratory and in the air would solve the technical problems; and nothing but a solution seemed to him to be rather insuperable if they went beyond a certain size. In the of the scientific problems could make air transport self-supporting. Everyone would four-engined machine made by his firm, which weighed 30,000 lbs., it seemed that agree that that was the solution to be obtained. It was deplorable to realise that the they were gradually approaching the point where materials had been used to the authorities seemed so little capable of realising this, that they allowed the technical best possible advantage, and there seemed little possibility of improving the use and research expenditure to be cut down with a view to national economy; he felt of those materials and getting over the theoretical disadvantage that was attendant very strongly that this was simply national extravagance . It was killing that which upon increased span. He did not know whether M. Breguet, with his great engineer- alone could economically keep the aircraft factories in being in the state neces- ing skill, had thought of any method by which that disability might be overcome. sary to provide the fighting machines for aerial defence—Britain’s, and indeed any Without a doubt, were it possible to obtain a scale effect with larger sized machines, country’s, first line of defence. by which they could fly with greater loadings per square foot and thus decrease Mr. F. HANDLEY PAGE said it was very interesting to have M. Breguet to the span, such an improvement might be possible. In that direction it was rather address a meeting of the Royal Aeronautical Society. He had been one of the great interesting to know that in nature some of the bigger birds which, he was assured by pioneers of aviation, and one who had devoted himself not merely to the type of Dr. Hankin, flew at the same speed, carried a greater load per square foot, although aircraft that they saw flying today, but he believed some of his earliest efforts were their landing speed was the same. It would appear that nature had got over struc- in the direction of the helicopter, which, if it had not produced practical results, tural difficulties by introducing a scale effect. Whether it was possible to utilise that 112 Chapter 3: The Design Revolution Document 3-4 113 in a very much bigger way in the construction of large machines he did not know. Mr. W. O. MANNING said he was sure everybody appreciated the honour Perhaps M. Breguet could say something about it. He again thanked M. Breguet for M. Breguet had conferred upon them by reading his exceedingly interesting paper. his interesting paper, and congratulated him on having read it in English. Those who had followed aviation from the early days would remember the large, Captain GOODMAN CROUCH congratulated M. Breguet on his very inter- series of aeroplanes known as the Breguet type, and the brilliant engineering design esting and rather optimistic paper, and the Society on having the opportunity of that invariably characterised them. He endorsed what Captain Goodman Crouch listening to one of the earliest pioneer aviators of France, who was also one of her had said to the extent that, although he was not connected with commercial avia- greatest engineers. tion, he certainly expected that the life of aeroplanes and engines today in commer- He recalled that he had had the honour of being associated with M. Breguet in cial work was considerably longer than 200 hours. He was not an engine builder aviation some 16 years ago, when he (M. Breguet) put into the air a type of aero- and could, perhaps, speak with less bias on that particular point, but he knew he plane which was called a double monoplane, since it was so unlike the usual box could introduce M. Breguet to one or two English engine builders who could beat kite form of biplane then known. People looked at this curious beast and asked what that performance. It would give them a great deal of pleasure when they remem- it was, since with biplane wings it had a fuselage and was almost entirely of metal bered that in the early days of aviation practically the whole of English aviation was construction. One had only to consider the machines of a few years later to realise dependent upon French engines if they were able now to return the compliment. that M. Breguet was a true prophet, for it would be remembered that he alone at M. Breguet had done a very great service to aviation by pointing out the enormous that time was the designer working on those lines. importance of improvement in aerodynamical efficiency. One tried to improve the With regard to the air transport costs quoted by M. Breguet, the list of prices efficiency of the machines, but one did not appreciate, until it had been pointed out shown for running costs of an aerial line of 800 km. length included an item of so clearly, what a very important matter aerodynamical efficiency was, not only in Frs.6.50 for sinking fund out of a total of Frs. 17.41. This, in his opinion, was a the saving of engine power, but in increasing the useful load and in cheapening the remarkably high percentage. It indicated that in the hypothetical case M. Breguet cost of the machine. With regard to M. Breguet’s large machine, he was, of course, had taken he had chosen a machine presumably with a far shorter life than that of up against the dimensional law that the weight went up considerably faster than the present-day aircraft, and he believed this item could be considerably reduced. This area and that there is a definite limit of size for machines of present-day construc- appeared to be borne out by the length of life assumed by M. Breguet when consid- tion and design. But it by no means followed that the limit was the same for other ering the advantages resulting from such improvements in engines and in mechani- types. It was possible that a very large monoplane, such as that referred to by the cal parts, as could bring reduction in sinking fund and upkeep charges. Under this Lecturer, presumably with highlift wings, and with the passengers, fuel, engines, head M. Breguet had assumed the life of present-day machines to be 200 or 250 etc., distributed along the wings, might be capable of being constructed for a rea- hours, and predicted the possibility of improving this to 1,000 hours. Captain Crouch ventured to suggest that even during the war the life of 250 sonable weight. He again thanked M. Breguet for his interesting paper. flying hours had been reached, and even surpassed by some of the heavy bombers, Colonel W. D. BEATTY joined with previous speakers in congratulating M. and the figure of 1,000 hours was hardly a prophecy since a total of approximately Breguet upon having read his paper in English. There was one point which had 800 hours had already been reached by a machine on the London-Paris Service. struck him, and that was in regard to the symbols used. The efficiency factor he had With regard to the advantages hoped for from the betterment of aerodynamic assumed corresponded to our English L/D. Although there were exceptional cases qualities, M. Breguet’s fineness coefficient indicated a figure for L/D of approxi- where the technical experts in the two countries did speak each other’s languages, mately 18~. This appeared to be optimistic, but he dare not say much in criticism the majority did not, but it was obviously desirable that they should all think in the of the point, since M. Breguet had already shown himself to be a true prophet. same mathematical language. With regard to that he was glad to, say that prelimi- Recent aerodynamic tests on a model of a Woyevodski type had given a maximum nary steps had already been taken, in conjunction with the French Air Ministry, L/D of about 12, and that, as far as he knew, was the highest figure yet reached for in order to get down to the same basis in England and France. While he did not a complete model. wish to plunge into the argument as to the possibility or otherwise of the very large M. Breguet’s remarks concerning future large machines were teeming with machine, he felt rather glad that there seemed to be such hope for the future. We interest, but he felt that M. Breguet was extraordinarily optimistic in assuming that still had a lot to do in the way of building up the traffic necessary to make it com- the increase, in volume available for passengers would be so much greater than the mercially practicable. increase in weight, power and cost. Mr. O. T. GNOSSPELIUS, after expressing his interest in the Paper, congratu Finally, he again thanked M. Breguet for the lecture, which from a technical lated M. Breguet upon his courage in assuming that what we called L/D, or effi point of view gave so much food for thought. ciency of the machine, could be improved, because he was quite sure, by his own 114 Chapter 3: The Design Revolution Document 3-4 115

experiments, that it could be done. They knew that certain experiments had been attain running efficiencies which would be considerably better than the 220 hours done at the N.P.L. and other places, and they got certain results, but if one made of the French engines. experiments one’s self one got quite a different outlook. He was sure M. Breguet was He also endorsed M. Breguet’s remarks as to aviation being the greatest future quite right in saying he could get something like 15 to 1 L/D, because he himself means of international communication, and in conclusion took the opportunity had made pieces of wood in the shape of bodies, wings and tails which gave that to thank M. Breguet for the courtesy which he had extended to him (the speaker) effect, so that he did not see why the complete machine should not do the same. when he went through his works. The trouble was that we did not know much about aerodynamics, and did not The CHAIRMAN translated some of M. Breguet’s remarks, in which he know the proper shape to make. It seemed to him that work on that line, was very explained that he was once of Mr. Handley Page’s opinion that six tons was the essential, because 151bs. per h.p. was not practicable; we wanted to turn it into 30. limit at which the weight grew so fast that the area could not be expected to bear In present machines the figure was more or less 151bs., and he was very glad that it profitably. He had proposed a complete departure in the type of construction, M. Breguet had brought forward this sort of figure for efficiency. which, although it had not yet actually been put into being, he thought would get Mr. A. P. THURSTON said that the lecture impressed one very much indeed, rid of the difficulty of the relation of weight to wing surface. He did not claim as an that famous men, like science, were international. We (in England) more or less invention at all, but he had made calculations by which, using the thick wing type regarded M. Breguet as one of ourselves, and felt it a very great honour to receive of machine, burying the engine and load in the wings, and distributing them care- a lecture from him. He was a pioneer of many things, but it was not realised that fully along the wings, on a 50-ton machine, he hoped to be able to arrive at much over 1,000 of M. Breguet’s all-metal machines were actually used in the fighting line the same wing loading and power loading as would be obtained on a smaller craft of during the war. That was a very considerable achievement. 2 or 3 tons. The wings would be 7ft. thick. M. Breguet had brought out very clearly that to increase the efficiency of a A hearty vote of thanks to M. Breguet concluded the proceedings. machine it was necessary to reduce the “useless surface” (surface invisible) to the minimum amount, and the ingenuity of our designers must be utilised in taking, off NOTES ON M. BREGUET’S PAPER. all extraneous corners and everything which caused waste by increasing that surface. Contributed by Captain W. H. SAYERS: I think it is most important and most But there was a point which was not, perhaps, brought out quite so clearly, and that encouraging to hear so very high an authority as M. Breguet expressing the opinion was that the efficiency in carrying weight per distance could be increased by actually that very great improvements in the aerodynamic qualities of the aeroplane are not increasing the speed of the machine and decreasing the lifting area. The great dif- merely desirable, but are also possible. The data given in his Paper are conclusive ficulty in this connection was that of landing speed. But there were ways of doing it proof—if proof be needed—of the value of any great improvement in the “fine- which would enable them to get a higher speed still in the air with smaller surface, ness”—or in usual English terms—the L/D ratio of aeroplanes. and yet maintain slow landing speeds. In other words, it was possible, as suggested The figures as to costs, etc., given in the Paper relate to French practice and by Mr. Handley Page, to get something of a scale effect by taking advantage—he are not directly applicable to British conditions. In certain respects I think British was not at liberty to say how—of certain properties of the air. He did think that it aircraft constructors may rightly claim that they can improve on those figures both would be possible to increase the present efficiency of our machines in order to get aerodynamically and in the equally important matters of the durability and lon- a greater weight mileage for a certain expenditure either of money or of fuel, and in gevity of their aircraft. Such criticisms do not substantially affect the justice of the that way increase the possibilities of commercial aviation. He had once taken the author’s general conclusions, with which I am in entire agreement. trouble to go through the figures, taking a line to India. Assuming there were a large I think, however, that the time has now come to question what have hitherto number of machines, taking the cost of maintenance of grounds, and petrol at 2s. been regarded as the fundamental bases of aeroplane design-bases which are appar- per gallon, assuming the organisation was so perfect that each machine could be ently accepted by M. Breguet for they are implied in his two equations Nos. 1 and 2. flown for 10 hours a day, and each machine would last on an average two years, on The assumptions are that an aeroplane may be regarded as a heterogeneous that basis he had made out that it was quite possible to maintain a good dividend assortment of surfaces and bodies, and that the forces on each of the component and charge passengers at the rate of 3d. per mile. surfaces or bodies, taken separately, may be added up and will then represent the With regard to M. Breguet’s remarks about engines, he would like to endorse total of the forces on the aeroplane. Every aeroplane designer knows that these what was said by Mr. Manning. It was only a few days ago, at Croydon, that a assumptions are in fact inaccurate. No isolated body of good form can be cut in Napier Lion was pointed out which had been running continuously for 450 hours two and have its resistance determined by summing the resistance or the forces on without being taken down. It was possible for British engines to do better still, and the parts. 116 Chapter 3: The Design Revolution Document 3-4 117

That the resistance and lift of present aeroplanes can be determined with rea- tional transport service will have to be of the amphibian type, and they will thus be sonable accuracy by this method is, I submit, evidence that present-day aircraft are able to utilise sea routes and have at their disposal the whole existing organisation in aerodynamically merely a collection of uncoordinated components. They will con- the big seaports throughout the world. tinue in this state for so long as designers allow themselves to be limited by a basis From carefully carried out experiments made in various laboratories I am able to of design which is entirely empirical and seriously misleading. It is not true that a state definitely that with the very good coefficients I quoted for “fineness” (indicat- given wing has definite lift and drag coefficients at definite angles of attack which ing L/D efficiency), propeller efficiency, and minimum h.p. will certainly be realised are independent of the body, tail and wing bracing structure to which the wing is in the near future. attached. It is equally not true that the body, tail and other organs which form the I would like particularly to reply to Mr. Handley Page on the subject of large complement to wings have force and resistance coefficients which are independent aircraft of the future. I can share his opinion that if aircraft are built on a larger of the wing. scale whilst remaining geometrically similar, the ratio of wing structure weight to Because existing aeroplanes are so bad that they behave nearly as though these total weight will increase, since, all other things being equal, the weight of the wing false assumptions were actually true, the designer comes to believe in them. Because structure, i.e., spars, ribs, interplane struts, bracing, etc., increases as the surface to he has come to believe in them, he has also come to believe that it is not possible power 3/2. For this reason I was for some time led to think that very large aircraft very greatly to improve the aerodynamic efficiency of existing types of aircraft. would not give results of interest, and that no comparison was possible between I am personally firmly convinced if designers can only be persuaded to forget all boats and aircraft, for even given an assured advantage to be gained by increasing the about the itemised resistances of the components with which they at present deal, tonnage of ships, rather the opposite would obtain in increasing the size of aircraft. and will regard a projected aeroplane as a single aerodynamic body, and will design At the beginning of last year, however, I started to give serious thought to a type of it with an eye to its lines as a whole—just as they would design, say, an airship monoplane with wings of a section deep enough to house engines, tanks and passen- body—that eye will very speedily be found possible to design a complete aeroplane gers. Further, by suitably distributing the loads on the wings it is easy to imagine an having a L/D ratio of 20/1 or over, or—in Nil. Breguet’s terms—with a “fineness” aircraft in which the dead weight, instead of increasing as the power 3/2 of the area, of .05 or less. will only increase directly proportional to the area, and under these circumstances there were no longer any disadvantages in increasing the size of the machines. TRANSLATION OF LETTER FROM MONSIEUR LOUIS BREGUET, DAT- I sketched on the blackboard after the discussion how I envisaged such an air- ED 28th APRIL, 1922, TO THE ROYAL AERONAUTICAL SOCIETY. craft. It would comprise three fuselages, one in the centre for the crew, controls, Gentlemen, —I return herewith draft copy of the report of the meeting of the instruments, etc., while the other two on the right and left respectively of the first 6th instant, and regret that I have been unable, owing to absence, to reply earlier, mentioned, and far enough apart, would be used as hulls or floats. These would as promised, to the various speakers who took part in the discussion. I very much contain first class cabins. Finally, the tanks and the float would be distributed inside appreciate their remarks, for which I thank them. the wings. This distribution in weights in large aircraft would obviously give them a Replying in the first place to Mr. Handley Page, I agree with him that the Paper considerable moment of inertia, but as there would be no need for “stunting,” their had in view only a part of the big problem of the future of aerial transport, but it is large lateral moment of inertia would mean a high degree of stability in the air. quite certain that while it is necessary to have good aircraft, it is equally indispens- With big monoplanes the reduction of parasite resistance can be pushed as far as able to have as perfect a ground organisation as possible, otherwise the use of aircraft one likes, and it is to be hoped that results similar to those of plain wings furnished will be very uncertain, however perfect they themselves may be. with the tail unit can eventually be reached. It follows that these large machines For instance, let us imagine modern navigation carried out as is actually the case would eventually have characteristics comparable with those of large birds, and the with superb boats, but lacking any organisation of ports, routes, provision of buoys, coefficient which would be applicable to them in that case will certainly be better lighthouses, wireless telegraphy, meteorological service, current charges, etc. The than those I quoted in the lecture. result would be practically negligible and its existence very precarious. To Captain Goodman Crouch I admit that certain British engines have a much I did not raise this question at the meeting as I am of opinion that if from, the longer life than 250 hours, but I ought to say that the average life of the engines present time until the fact is accomplished, sufficient funds are available, it ascertain used by French companies does not reach this figure. that excellent aerial ports, which are more easily and much more cheaply established I am entirely in agreement with Captain Goodman Crouch that in a few years than seaports, will rapidly come into being. aeroplanes will obtain an average life of 1,000 hours, and for this reason I have every It is foreseen, however, that the aircraft destined to maintain the big interna- hope of seeing engines reach 2,000 hours. 118 Chapter 3: The Design Revolution

With regard to Mr. Thurston’s remarks, may I say that actually there were more than 6,000 metal machines, and not 1,000, put into service during the war which are still being used. Finally, I would like to confirm what was the basic point of my paper, i.e., that technical research should be pushed forward without relaxation so that improvements already envisaged may be realised with as little delay as possible as well as those for which as yet one hardly dares to hope.

Yours, etc., (Signed) LOUIS BREGUET. Document 3-5 119

Document 3-5

Captain W. H. Sayers, “The Lesson of the Schneider Cup Race,” Aviation 15 (5 November 1923): 577-580.

It was clear to many informed observers of the American victory in the Sch- neider Trophy competition of 1923 held in England that the key to the American success lay in how the designers of the winning Navy Curtiss CR3 aircraft had found ways to limit the amount of what was at the time generally called aerodynamic “resistance,” meaning drag. In this short commentary that first appeared in the British magazine The Aeroplane, British RAF captain W. H. Sayers emphasized the “very complete way in which the whole detail design [of the Navy Curtiss racer] has been studied throughout with the idea of keeping resistance down to the absolute minimum.” Primarily a critique admonishing the British Air Ministry for not providing the type of funds the U.S. government had been giving to winning such international air competitions, Sayers’s editorial reported that the American emphasis on “detail design” for the purpose of reducing drag had made all the difference in the recent Schneider Cup race. The Americans had made every effort to make their airplane sleeker and faster. As a result, they had produced an integrated system designed for one purpose: speed. Aviation historians still debate how influential the air races actually were on the aircraft design revolution of the late 1920s and 1930s and the more sophisticated civil and commercial aircraft that emerged from it. But there is no doubt that the U.S. government’s discontinuation of formal sponsorship of international racing efforts left a void and somewhat slowed the pace of experimenting with cutting-edge aeronautical technology.

Document 3-5, Captain W. H. Sayers, “The Lesson of the Schneider Cup Race,” Aviation 15 (5 November 1923): 577-580.

The Lesson of the Schneider Cup Race “America Won Because it Entered the Most Perfect Examples Of Racing Air- craft Yet Seen in Europe”

Under the above title W. H. Sayers, (Captain, late Royal Air Force) contributed a very interesting article to our English contemporary The Aeroplane. It is reproduced here not only because of its evident technical value but also because of the clean sportsman- like manner in which it appreciates the victory of our entries in the Schneider Cup race. – EDITOR. 120 Chapter 3: The Design Revolution Document 3-5 121

The American team won the Schneider Cup because they entered for it what In making this statement one does not refer specifically to its low weight per hp. are undoubtedly the most perfect examples of racing aircraft that have yet been or its obviously excellent running qualities. The engine is astonishing--considering seen in Europe. It is entirely beside the point to suggest that they won the race on its great output--on account of its extremely small frontal area and of the absence machines that would not have stood up to the Navigability Tests in rough weather. of any projecting gadgets for which it would be necessary to provide bulges in the They stood up to the tests in the weather that actually befell them, and their cowling. behavior on the water was of a nature which somewhat upset one’s preconceived There are other engines which can cheerfully face comparison with the Curtiss ideas as to their seaworthiness. Their floats are one believes distinctly heavy, they are on a weight per horsepower and a reliability basis. One does not know of one of any- certainly extremely strong. thing like the power which would go into the Curtiss racing fuselage with absolutely nothing sticking out except the airscrew boss and the stub ends of the exhaust pipes. The Cleanest Things Ever Seen Personally one has no doubt that a big body with nothing whatever sticking out is As for the rest of the machines they are the cleanest things one has ever seen. better for speed work than is a smaller one with an assortment of minor bulges to The type 29 racing Nieuport biplane was one imagined as clean as a biplane could accommodate odd excrescences which will not go inside the main lines. be. The Curtiss CR3 is a little cleaner in the matter of minor details such as wire fit- The Curtis body is small--and it has not one single bump on it from nose to tail. tings, etc., it has no radiator other than the surfaces of the upper plane, and despite And it is pretty certain that a small clean body is better than a large clean one when the fact that it has an engine of some 500 bhp as against the Nieuport’s 320 bhp, one is trying to push it along at 200 mi./hr. or so. that engine is stowed away in a fuselage which is certainly of no greater--and looks For those who contemplate the designing of a machine to beat the Americans to be of distinctly less--cross-sectional area than that of the Nieuport. in the next race for the Schneider Cup it may be as well to remark that it is alleged So far as the general arrangement of the machine is concerned the Curtiss that the new Wright T3 engine of 700 hp. has an even better frontal-area-to-power machines show no surprising features. They use a normal type of thin wing--one ratio than the Curtiss D12, and the reports that the Pulitzer Trophy machines with believes it to be a “Sloan” section--which looks not unlike a RAF15. There is noth- this engine have achieved speeds of within a mile or two of 250 to the hour seems ing abnormal about the wing wiring or strutting. The gap/chord ratio is not far to be entirely trustworthy. No English engine maker has been able to afford to design an engine specially from the usual value of 1. for racing purposes, and to that extent we are necessarily handicapped over this The fuselage is an exceedingly pretty veneer--built affair-made one believes in matter of body size. two halves--and the tail unit has no striking peculiarities. The floats are of a type which is now practically standardized in America for all The Reed Airscrew float seaplanes. They are of the long type, designed to be stable fore and aft without a tail float, have one shallow step close under the C.G. of the machine, with a very Another factor contributing toward the success of the American machines is the long and fine tail behind it. The bottoms have a fairly steep Vee throughout, and the Reed duralumin airscrew. Those used in the race were 8 ft. 9 in. diameter and they were directly driven at 2,300 rpm, giving a tip speed of 1,054 ft. per sec., which was section above the bottom is practically semicircular throughout. near as no matter the velocity of sound. According to the R.A.E. the efficiency of an In the matter of air resistance they are undoubtedly good. On the water--as already remarked--they are surprisingly amenable to the wishes of the U.S. Navy’s airscrew with a tip speed equal to the speed of sound is as nearly as can be zero. undoubtedly able pilots. The Reed airscrew at this tip speed is obviously fairly efficient--and it allows But it is not the general design of the major components, or the lay-out of the the Americans to dispense with a reduction gear on their racing engines. They can thus use a lighter engine--a smaller diameter airscrew,--and keep their undercar- whole machine which accounts for the surprising performance which the Curtiss machines undoubtedly possess. This is due to the very complete way in which the riage struts of reasonable length and still have water clearance. All these little things whole detail design has been studied throughout with the idea of keeping resistance count. A special article on the Reed airscrew will be published as soon as possible. down to the absolute minimum. The engines, radiators, and airscrews are the most striking features of the Amer- ican machines. According to our authorities to attempt to run an airscrew of this The Curtiss Engine diameter at this speed is to invite complete failure. The Americans have shown that our authorities are wrong on this point, and it is now up to us to find out how it is The Curtiss D12 engine must be considered to be a most important factor in done. the whole. It is evidently an engine of the very highest class--but more than this it The wing radiators are pretty certainly a very important feature. To all intents is definitely a racing engine such as does not exist at present outside of America. and purposes using this type of cooling system is equivalent to getting one’s cool- 122 Chapter 3: The Design Revolution Document 3-5 123 ing for nothing whereas with a normal radiator the power absorbed in pushing the in order to win the Schneider Cup, whereas the British Aircraft Industry had no radiator of a 500 hp. engine at 180 mi./hr. would not, be much less than 5 hp. money to spare to defend it. There is no novelty about the idea of using wing or body surface for cooling. There are obvious practical difficulties, and for some purposes some disadvantages. There The Real Cost of Winning is no disadvantage, for racing purposes at any rate, which can counterbalance an This explanation is accurate enough in its way. It is not however the whole increase of at least 10 percent in the horsepower available for propulsion. explanation. In the first place it has to be remembered that what has been spent directly for the purpose of financing the Schneider Cup raid is a very small propor- Detailed Design tion of what has had to be spent to make that raid possible. The Curtiss D12 engine These three main items--all concerned with the power plant--would by them- is a direct descendant of the Curtiss-Kirkham engine produced just after the war. selves give the American team ample excuses for having defeated us in the race. Between that engine and the present there are at least three quite distinct varieties Beyond them however there is still something to be learned. The designer of the of Curtiss 12-cylinder high-performance engine each one an advance toward the Curtiss racer was not content with general cleanness of design. He has seen to it ideal racing engine. No English manufacturer has been able to carry on the inten- that no unconsidered trifle was forgotten in the original lay-out and was afterward sive development of high performance engines to this extent since the war although permitted to cause a minor irregularity of outline. As a matter of fact there are one the engine work actually accomplished in this country under much more severe or two small fittings for bracing wires on the underside of the top wing which look limitations on expenditure has possibly been of greater practical utility than that as though they might have been more carefully kept out of the draught. They are accomplished in America. confined to the undersurface, where they have the least effect. It is extremely diffi- Over and above engines it may be recollected that since the War the American cult to make a satisfactory wire terminal inside the wing surface--doubtless the wing Aircraft Industry has produced a new batch of special racing machines every year- radiators add to the difficulty--and there is precious little of these fittings anyway. -first of all for in attempt on the Gordon-Bennett trophy and since then for their Otherwise the machines are as nearly perfect in this respect as they could be. own Pulitzer Trophy races. Last year alone America built more racing machines than The ailerons--on the bottom wing only--are recessed into the wing so that one could Great Britain has built in her whole history. Thus the actual expenditure which not put a strip of thin paper through the gap. Rudder to fin and elevator to tail- has made it possible to produce the Schneider Cup Racers is extremely large--and plane hinge gaps are covered with rubber sheeting so that there is no gap in fact, and almost the whole of it has been provided by the Government of the United States. there is not a projecting control lever or wire anywhere on the machines. But it may be doubted whether the United States despite its large expenditure The gasoline tanks are in the floats. The hand holes which give access to the on this special object has an Aircraft Industry which is financially in a position to filler caps, and those for inspecting the float interiors are recessed slightly into the overwhelm the French Aircraft Industry in the production of racing aircraft. The surface. After these tanks were filled they were covered neatly by a square patch of Americans beat England--and still more thoroughly France--in the Schneider Cup doped on fabric, varnished as perfectly as the rest of the float and giving an unbro- race because regarded as racing machines pure and simple their Curtiss CR3s were ken surface. as nearly perfect in every detail as they could be made. The undercarriage struts go clean into the floats--apparently they are built in before the float is planked--and there are no fittings disfiguring their exposed sur- What is Needed to Regain the Cup face. And so on throughout. The British Aircraft Industry can produce machines at least the equal of the CR3s. It needs for this purpose a fair amount of money--which at present it does High Class Workmanship not possess--but it need not expend a tithe of the money which has been spent on Not only is the workmanship and exterior finish of a very high class but it is this purpose in America during the past five years. For there is not the least doubt obvious that the designer has taken as seriously the polishing up the last detail of his that so far as general technical ability is concerned the British Aircraft Industry is design as he took over the general lay-out. In fact he must have taken considerably ahead of the American, and that much that America has learned in the process of more. developing racing aircraft we have learned from our general experience. The merits of the American racers have been dealt with at this considerable But we have to learn that racing machines cannot safely be improvised at the length in order to make it quite clear how the Americans have achieved their success. last minute. The perfection of detail, the avoidance of every unnecessary compro- It has been very generally asserted that the explanation is simply that the American mise are essentials in the design of a successful racing airplane. Government was willing to spend as much money as might prove to be necessary In aircraft built for either war or commerce design throughout is a matter of 124 Chapter 3: The Design Revolution Document 3-5 125 compromise between the conflicting requirements of aerodynamic efficiency prac- hull a bit dirtier on the water but correspondingly reduces the air resistance. tical utility, and cost, and British designers who have been brought up strictly to Fairing pieces have been added behind the side of the main step, and wing- design for practical utility seem sometimes inclined to under-estimate the aerody- tip floats of an elliptical cross section, of reduced buoyancy-- but fitted with bow namic importance of detail design on the resistance of the complete machine. hydrovanes to give dynamic support on the water have been fitted. Taking everything into consideration one has no doubt whatever that Great The engine cowling has been cleaned up appreciably and a smaller radiator of Britain can bring the Schneider Cup back from America--if not next year at least the the “long tube” type has been fitted. Altogether the effectiveness of the modifica- year following. If we are to do it next year we have to start work at once, and some tions is shown by the increase of some 10 mi./hr. in the average speed attained this money has to be raised quickly. year as compared to the last race for the Cup. In fact the “Sea Lion” did astonishingly well--the American racers ought to have Inter-Governmental Sports? beaten her by more than 20 mi./hr. if their superior cleanliness of line had been all But on the whole the question of whether we are to try for the Cup again seems gain, and the result actually achieved would seem to indicate that the flying boat to be a matter for the Air Ministry to settle up with the Treasury. One had an idea type has intrinsic aerodynamic advantages as against the twin-float seaplane. that the Schneider Cup was instituted as a sporting Trophy. Since the advent of the Personally one believes that the “Sea Lion” with wing radiators and a totally mechanically-propelled vehicle one has had to modify one’s ideas on the subject of cowled-in “Lion” could give the CR3s a very good race indeed without further the connection between sport and commerce far as motor car and airplane racing alteration. Unfortunately we shall not again have CR3s with which to compete. is concerned. But the advent of the U. S. Government as a direct competitor in a “sporting” The CAMS event is somewhat of a startling innovation. Obviously if the American Government The two CAMS boats are of very similar design--the only important differ- has decided that the development of racing aircraft will lead to technical develop- ence being that the type 36--designed for last year’s race is a tractor with the pilot ments of real utility it is perfectly entitled to finance the design and building of such behind the wings, whereas the 38 has a pusher airscrew with the pilot ahead. Both craft and to enter them for International events. And certainly none can suggest that are extremely clean and taking-looking boats. The type 33 is alleged to be some 10 in any detail the American team at Cowes and their auxiliaries have shown them- mi./hr. the faster of the two. selves to be anything other than sportsmen in the highest sense of the term. But it The hulls are thoroughly well-made boat-built jobs, roughly rectangular in sec- is pretty obvious that no private individual or corporation can afford to compete in tion forward, with a domed upper surface and the concave bottom which is favored such a test with the Government of the richest Power in the World. by Signor Conflenti and by Mr. Short. They have but one step, the tail of the hull Personally one believes that it would be all to the good were the precedent set rising fairly steeply behind it, and the fin and tail plane are built up as part of the by the U. S. Government in this matter followed by others. Inter-Governmental hull. sporting events should have the happiest results on International relations if they The wings are of the normal single-bay biplane type, but with a smaller chord on the lower wing. The engine nacelle is built up into the upper center section and were to become normal events but one cannot expect the British Aircraft Industry is very thoroughly streamlined--the airscrew in both tractor and pusher types being to compete on level terms with the American Government. fitted with a large spinner. If the Air Ministry agrees with America as to the importance of racing as a method of developing the technique of aircraft design, then it is up to it to accept The engine mounting is used as the central support of the upper wing, thus the perfectly open and honest challenge which America has offered, and to enter securing a minimum of strutting at this position. a team for the Schneider Cup and the Pulitzer Trophy races either next year or the Both boats are exceptionally clean and given engines of the same class as those year following. against which they had to compete should have made a very good show. One believes that in regard to seaworthiness these two boats are distinctly good. The “Sea Lion” They make a lot of spray getting off, and if they do not come away cleanly they fall back with a somewhat startling thud. But they probably unstick better in a bit of The general characteristics of the “Sea Lion” are already pretty well known, because the machine is actually that which won the Schneider Cup last year at a sea than they do in a flat calm. They are--so far as one can judge--rather better Venice. But certain modifications of detail have been made. The nose of the hull has rough-water craft than the American float machines, and closer approaches to pure been modified--the marked “ram” of recent Supermarine hulls has been consider- racing machines than is the “Sea Lion.” They were however outclassed in the matter ably reduced by bringing the chines in more gradually which probably makes the of engine and in addition their luck was right out. 126 Chapter 3: The Design Revolution

The Latham The big Latham boat with Lorraine engines in tandem was the foul-weather hope of the French team. Her flight to Cowes in a full vale was a very fine performance indeed, but one cannot feel very greatly enthused over the design. There is no doubt that for really rough seas one needs a heavily-built seaplane with plenty of engine power. The Latham had these qualifications and perhaps with the limited choice of engines available in France it may be regarded as a very fairly satisfactory attempt at a powerful and reasonably fast seagoing flying boat.

The Blackburn Pellet The fate of the Blackburn “Pellet” is very greatly to be regretted. All those of the Blackburn Staff and of Mr. Saunders’ staff who worked so hard to get the machine ready in time for the race deserve every sympathy with the very bad fortune that befell their efforts. One sympathizes with them the more because in many respects the “Pellet” was of a distinctly taking design, and because one feels sure that had there been a reasonable opportunity of getting the machine tried out before the race it might have put up a really good performance As it was the machine had no chance. It was designed to make use of an existing hull which in fact proved to be defective in strength and unpleasant on the water under conditions quite different from those for which it was originally designed. Nevertheless one believes that it could have been made to get off and one is certain that in the air it would have been very fast. But it was an effort to improvise a racing machine--and since the Americans have taken a hand in the game improvisations have a precious poor chance in Inter- national Air Races. Mr. Kenworthy’s lack of seaplane experience spoiled what little chance remained. The hull itself was of the Linton Hope type in general design, and of very clean lines. It was however intended for a much less powerful machine than the “Pellet” and was pretty certainly overloaded on the water. On this hull was mounted a bottom wing of small chord and span, and above it--supported on a steel tube cabane and outwardly-raking N struts a top wing of considerably larger dimensions. The Napier “Lion” was mounted above the top wing over the central cabane, and drove a tractor Leitner-Watts steel airscrew. Originally tubular radiators recessed into the lower surface of the top wing were fitted. These were not satisfactory, and a single Lamblin, mounted below the engine was later fitted instead. The machine was undoubtedly distinctly tricky to handle on the water. A hull of relatively small beam--a high C.G., a narrow base for the wing-tip floats, and the torque of the Napier “Lion” made her very prone to immerse a wing-tip on opening up the engine. The hull obviously tended to porpoise badly at low speeds, and this combination of qualities needed a lot of handling. Document 3-6 127

Document 3-6

John K. Northrop, quoted in Garry R. Pape, Northrop Flying Wings: A History of Jack Northrop’s Visionary Aircraft (Atglen, Penn.: Schiffer, 1995).

The following comments by Jack Northrop, quoted in a 1995 book detailing his aircraft designs, reflect his affinity for streamlining, but from the viewpoint of his extraordinarily innovative approach to aircraft construction technology. In this fashion, Northrop’s abilities to integrate a new type of structure accentuated the aerodynamic capabilities of new airplane design. First tried on the diminutive Lock- heed S-1 sport plane of 1919, Northrop’s technique of forming a plywood fuselage between a concrete mold and an inflatable rubber bag enabled a manufacturer to build an exceptionally streamlined airplane, and to do so rather economically. As for the wing, Northrop looked for inspiration to Dutch aircraft builder Anthony Fokker’s cantilever designs. He went with cantilever even though Lockheed president, Allan Loughead, was against it, because he felt that potential buyers would feel the plane was unsafe without an appropriate amount of bracing struts and wires. A highly intuitive thinker, Northrop reasoned that the streamlined combination of cantilever wing and monocoque fuselage, both without drag-producing struts and wires, could operate at less power and do a better job than aircraft that weighed much more. His gamble proved highly effective, and the new airplane, the Vega, became one of the breakthrough designs of all time. 128 Chapter 3: The Design Revolution

Document 3-6, John K. Northrop, quoted in Garry R. Pape, Northrop Flying Wings: A History of Jack Northrop’s Visionary Aircraft (Atglen, Penn.: Schiffer, 1995)

NorthrOp Flying Wings Quotes from Jack Northrop’s First Company: A steady program of development and refinement has been underway for the past twenty years until we have at present [1930] a number of carefully designed and comparatively efficient planes embodying streamline fuselages, carefully cowled engines, and ‘clean’ landing gears with superfluous struts, wires and fittings sup- pressed to an absolute minimum. It seems quite apparent that our best designs are close to the limit of practical efficiency; yet we find that their maximum over-all L/D (lift-drag) ratio is only about 10, whereas the L/D ratio of the active supporting surfaces of an airplane is normally double this amount. An analysis of the items adding to parasite drag in the normal design shows that landing gear, power plant, fuselage, interference and bracing, and control surfaces are the major contributors to parasite power loss; the item of control surfaces being by far the smallest. Individual examination of the various units shows that nearly all possible improvement has been made in existing designs.

“We didn’t dare to go the whole way and eliminate the tail.”

It was obvious that the ideal flying wing was impractical or impossible to construct except in very large sizes. The airplane cannot justify its existence unless it has capacity to carry a comparatively bulky cargo. With conventional wing thicknesses it was impossible, even with very large wing taper, to build a flying wing in sizes of less than 150 or 200 foot span.

The whole objective was to build as clean an airplane as we could possibly conceive in those days. The average airplane had struts or wires or fuselage forms that weren’t as smooth or streamlined—with as low a drag as possible. It was pretty obvious, it seemed to me, that a full cantilever wing neatly faired to the fuselage on a perfectly streamlined fuselage would take less power to do the job than some other types. So it was a breakthrough in that we went wholeheartedly into, for the time and at the time, to conceive as clean an airplane as we could. Fortunately, the work in Santa Barbara some years previously had already developed the technique for building a fuselage. It was then just the necessity of designing a full cantilever wing. Document 3-7 129

Document 3-7

B. Melville Jones, “The Streamline Aeroplane,” Aeronautical Journal 33 (1929): 358-385.

In 1929, Great Britain’s leading aerodynamicist, Cambridge professor Dr. B. Melville Jones, presented what in retrospect must be regarded as the single strongest statement made in favor of the advantages of reducing drag through streamlining that has ever been made in the history of aviation. The paper he read before the Royal Aeronautical Society, entitled “The Streamline Airplane,” outlined conclusively how streamlining directly reduced the total drag of an airplane. Jones stressed that such drag reduction would result in higher cruising speed and lower fuel consumption and would consequently increase range and payload. The Cambridge don pointed out the inefficiency of previous aircraft designs, and he cited reduction of form drag as the most important obstacle to increased aerodynamic efficiency. The aeronautical community received Jones’s landmark speech with tremendous enthusiasm, and their reaction to his talk indicated the first widespread conversion of an entire aeronautical community to the benefits of streamlining.

Document 3-7, B. Melville Jones, “The Streamline Aeroplane,” Aeronautical Journal 33 (1929): 358-385.

During the past two years Professor Jones has been turning his attention to quantitative work on control. He has recently completed a first stage in a research showing by instrumental records exactly what happens to all the controls and to the aeroplane itself for certain specified movements made by the pilot. The flights were carried out largely by Professor Jones in an aeroplane specially loaned by the Air Ministry to the Cambridge University Air Squadron; the results of this pioneer research work provided a basis for similar experiments on other aeroplanes to be carried out at the R.A.E. and at [Martingham]. About a year ago Professor Jones first drew the attention of the A.R.C. to the great step forward that was possible in the improvement of aeroplane performance by comparing the resistances of an actual aeroplane with those theoretically possible. Much further evidence has been added since that date with the consequence that the A.R.C. initiated a series of researches on the interference of one part of an aeroplane on another. The first results of the interference research, carried out by a sub-committee under his chairmanship, are now to hand and have justified the prediction made by Professor Jones as to the large improvement that was possible. Professor Jones has been an independent member of the A.R.C. since its appoint- ment in May, 1920, and has assisted in a number of further matters to which that committee has paid attention since that date. 130 Chapter 3: The Design Revolution Document 3-7 131

THE STREAMLINE AEROPLANE putting a theoretically leaky argument such as this before a body of scientists, who are of necessity more concerned with detecting fundamental errors in reasoning BY B. MELVILL JONES, A.F.C., M.A., F.R.AE.S. Ever since I first began to study Aeronautics I have been annoyed by the vast than in devising engineering approximations, you would understand why I have gap which has existed between the power actually expended on mechanical flight gone to some trouble to expose, as precisely as I am able, the assumptions upon and the power ultimately necessary for flight in a correctly shaped aeroplane. Every which the final conclusion is reached. year, during my summer holiday, this annoyance is aggravated by contemplating the My lecture therefore deals with the possible reduction of aerodynamic head effortless flight of the sea birds and the correlated phenomenon of the beauty and resistance, or drag, and hence of the power required for flight. At the outset it is grace of their forms. perhaps worth noticing that such reductions will not greatly reduce the maximum We all possess a more or less clear ideal of what an aeroplane should look like; power necessary for a given service, since maximum power is required for getting a kind of albatross with one or two pairs of wings—depending on whether we live off from the aerodrome—an operation in which weight is the predominating factor. in Germany or Britain. In our more sanguine moments we even—like Alice and the Reduction of drag will, however, enable an aeroplane of a given power loading, cat—see the wings without the albatross. But progress towards this ideal, so far as either to cruise at a higher speed or with a lower petrol consumption. This again the general purposes craft is concerned is, we must all admit, painfully slow. It has will result in increased range or paying load, both factors of the first importance in seemed to me that a contributory factor to the slowness of this evolution has been aeronautical development. the lack of any generally understood and easily visualized estimate of what could be achieved were the difficulties in the way of realizing the ideal form overcome. Before the power required to overcome drag can be interpreted in terms of There is a natural tendency to decide on one day that the gain—say 20 percent on engine power, the airscrew efficiency must be taken into account. Efficiencies as the total drag, or 7 percent on the speed—to be had by spending endless trouble high as 75 percent are practicable on present-day craft, and efficiencies higher than on improving the undercarriage design, is not worth the trouble; on the next day say 85 to 90 percent are unlikely to be achieved in the near future; hence the pos- to come to a similar conclusion about the drag of the engine cooling apparatus; on sible power economies to be obtained from further airscrew development—e.g., the next day about the wires, struts and minor excrescences; and on the next about through the use of the variable pitch screw—through by no means negligible, are the pilot’s view; omitting to notice that if all the improvements were made at once unlikely to be very great. I shall not deal with the airscrew problem further, except the total gain would not be some insignificant percentage of the whole, but might in so far as the body-airscrew interference may influence my main argument. reduce power consumption to a small fraction of its original value and so extend Thanks to the combined efforts of Lanchester and Prandtl we can now isolate the range and usefulness of the aeroplane into realms which would otherwise be from the whole power that part—the power to overcome induced drag—which is unattainable. expended in supporting the weight on a wing of finite span. This induced power, as Considerations such as these led me some time ago to examine the possibility of it may be called, depends primarily on “span loading” that is weight/(span)2. For a devising simple formulae for the power required by the ideal streamlined aeroplane, biplane of reasonable proportions the induced power per 1,000 lbs. weight can be and this lecture is a result of that examination. The very simple formula which I expressed as 1 shall discuss is not precise—the present state of aerodynamical knowledge does not 2.80 ϖ/σVm allow precision in this matter. It is not possible, for example, to say that lower power expenditures than my estimate might not occur. I am not so bold as to prophesy any where Vm is the speed, with one hundred miles per hour as the velocity unit. limit to progress, but I think that I have made out a case that the power expended σ is the ratio of the air density to that of standard air, and on the ideally streamlined aeroplane would not be greater than my estimate. Since ϖ is the span loading in pounds per square foot. this is of the order of one third the power expended on modern passenger carrying aeroplanes, the estimate should suffice, until such time as it more nearly represents The brake horse-power corresponding to the induced power is obtained by actual performance; by which time we may hope that aerodynamic science will have dividing by the airscrew efficiency. Assuming an airscrew efficiency of 75 percent, progressed to a stage at which more accurate estimates can be made. an average span loading of say 2.2 lbs. per sq. ft., and making σ equal to unity, we The formula at which I arrive is so simple and obvious that many will consider get the following table for the b.h.p. required for the induced power in a normal that I have wasted breath, ink and paper in putting it before you. That may be, but biplane. the difficulty lies in the choice of the quantities to be neglected and the argument here bristles with points where the scientist can pick holes. If you had had the job of B.h.p. per 1000lbs. weight 132 Chapter 3: The Design Revolution Document 3-7 133

Speed m.p.h. For Induced Drag total normally supplied (Ratio) Total Power exposed surfaces. In such a flow the pressures normal to the solid surfaces are very Induced Power closely equal to the pressures of the corresponding flow of the hypothetical fluid, but the action of the real fluid in the thin boundary layer causes additional tangen- 90 9.1 35 25% tial surface forces, or skin frictions, which are not present in the hypothetical flow. 100 8.2 45 18% An ideally streamline aeroplane may therefore be defined as one which:— 120 6.8 75 9% (a) Generates a flow identical, except in a very thin “boundary layer,” with the 150 5. 5120 5% flow of an inviscid fluid. (b) Experiences a pressure distribution identical with that due theoretically to Although the “induced power” is an important item in the power account at the the inviscid fluid; lower cruising speeds, it is not the predominating factor at speeds above 90 m.p.h. and therefore Moreover, since it is clear from theory that no notable reduction in induced power (c) Experiences a drag which is the sum of the induced drag and the tangential is to be obtained without using much larger wing spans, I shall not in this lecture or skin friction forces resolved in the down-wind direction. examine the problem of reducing it still further. Since no great power reductions are to be obtained either by improving airscrew Like all ideals, the ideally streamline aeroplane cannot exist; the boundary layer efficiency or by reducing “induced power,” it remains to examine the possibility of must have some thickness, so that the flow cannot be exactly the same as one in reducing that part of the head resistance which arises merely because the aeroplane which there is no boundary layer. If the “Reynolds’ number” is small, or in other is being dragged through the air, without reference to the fact that is must support words, if the effective viscosity of the fluid is high, the boundary layer becomes so its own weight. Using terms in common use, this part of the whole drag is described thick as to make the above definition meaningless, but with the very high Reynolds’ as the sum of the wing profile drag and the parasitic drags. The power required to numbers, typical of aeronautical practice, the approximation for a good streamline overcome it is seen in the table above to lie between 75 and 95 percent (according body is very close. to speed) of the total power applied to the modern general purposes aeroplane. Any It is, of course, well known that, unless bodies are carefully shaped, they do not serious reduction in this item will therefore have an important influence on the total necessarily generate streamline flow but shed streams of eddies from various parts head resistance. of their surface. The generation of these eddies, which are continually being carried We all realize that the way to reduce this item in the power account is to attend away in the air stream, requires the expenditure of power additional to that required very carefully to “streamlining.” It is the main object of this lecture to examine how to overcome induced drag and skin friction. The power absorbed by these eddies far this item can be reduced by perfect streamlining, and this incidentally will involve may be, and often is, many times greater than the sum of the powers absorbed by a preliminary explanation of what I mean by an ideally streamline aeroplane. skin friction and induced drag. The drag of a real aeroplane therefore exceeds the We recall first a proposition relating to that mythical and much abused sub- sum of the induced drag and skin friction drag by an amount which is a measure of stance, “the perfectly inviscid incompressible fluid.” Theoretically when this hypo- defective streamlining. thetical fluid streams steadily past a body, such as an aeroplane, the forces exerted Having arrived at the conclusion that the drag of the ideally streamline aero- on the surface of the body are everywhere normal to the surface; no tangential or plane is the sum of the induced and skin friction drags, the next step is to estimate skin friction forces can arise. The total reaction on the body, the resultant of all the magnitude of these drags. Methods of estimating the former are well known and the pressures acting normally to the surface, is such that the “drag,” or down-wind an approximate formula is given on page 359. No corresponding theory yet exists component, will be related to the “lift,” or cross-wind component, according to the for the estimation of skin friction on curved surfaces such as those of the wings and Prandtl theory of induced drag. In other words, Prandtl’s induced drag is the sum of body of an aeroplane. We are therefore forced to adopt some empirical method of the down-wind components of all the normal pressures which would be set up by investigation. The simplest experiments on skin friction are those upon a thin flat that steady flow of the hypothetical fluid which gives the required lift. It follows as plate edgewise to the wind, for here the resultant, in the down-wind direction, of a particular case of this proposition that the steady flow of the inviscid fluid which the surface pressures is necessarily zero. The best experiments, of which I am aware, gives no lift also gives no drag. upon the drag of such a plate at high Reynolds’ numbers were made at Göttingen In aeronautical nomenclature a “streamline body” is one about which the flow some years ago. They lead to the formula of a real fluid, such as air, approximates closely to a steady flow of the hypothetical − .15 inviscid fluid, except in a very thin layer called the “boundary layer,” surrounding the kF = 0.019 R 134 Chapter 3: The Design Revolution Document 3-7 135

2 where kF, the skin friction coefficient, stands for (skin friction)/ρV E and R stands for ρVl/µ where V stands for the relative velocity of air and plate. E stands for the total exposed area of both sides of the plate. l stands for the length of the plate in the wind direction. ρ and µ are the density and viscosity of the air. R is thus the Reynolds’ number involving the dimension parallel to the wind direction.

The experimental range was approximately from R = 2 X 105 to R = 107. This expression (1) relates to conditions in which the “boundary layer” is turbulent over the greater part of the plate, that is to say, the air very close to the surface of the plate is eddying—as it were, rolling on the surface. Burgers, of Delft, has shown that near the front of the plate the boundary layer is not turbulent, the separate layers of fluid near the surface of the plate slip over each other smoothly. When the Reynolds’ number is small this smooth region extends over the whole plate and the skin friction drag has, in these circumstances, been shown to agree closely with the theoretical expression

− ½ kF = 0.66 R

Thus at very low Reynolds’ numbers the drag coefficient is as equation (2), but at high numbers as equation (1). In Figs. 1 and 2 I have plotted curves representing the expressions (1) and (2), above, on the assumption of smooth and turbulent boundary layers respectively. The upper curve in each figure represents expression (1) and the lower expression (2). At medium Reynolds’ numbers, where the boundary layer is smooth over the front part of the plate and turbulent over the rear part, the average drag coefficient for the whole plate will lie between the upper and lower curves of Figs. 1 and 2; its exact value depending upon the distance from the leading edge at which the breakdown from laminar to turbulent flow occurs. The curve followed by the drag coefficient in this intermediate condition may be described as a “transition” curve, between the lower and upper curves of Figs 1 and 2. I have lightly dotted in roughly calculated transition curves, on the assumption that the critical change from smooth to turbulent boundary layer occurs when the Reynolds’ number formed from the distance of the breakdown point behind the leading edge has certain arbitrarily selected values, e.g., 105 and 5 X 105. Burgers has found that for the flat plate the value of the Reynolds’ number at this point of breakdown depends on the steadiness of the air in the tunnel before reaching the plate. When the flow was as steady as he could make it the change occurred when R = 5 X 105. When the flow was deliberately made unsteady, by bringing the wind tunnel honeycomb nearer to the plate or by placing wire gauze across the tunnel in front of the plate, the break up of the boundary layer occurred 136 Chapter 3: The Design Revolution Document 3-7 137 earlier, R = 105 approximately. Thus in the former case the drag coefficient of the flat but one set of points, those for thin wing sections R.A.F. 25 and 26 at values of R in plate should be expected to follow the lower curve until R has increased to 5 X 105 the neighborhood of 3 X 105, show a tendency to fall towards the lower curve. We and then follow the transition curve which starts at this point, to the upper curve. may summarize this diagram by the statements that:— In the latter case the change would take place along a transition curve which leaves When the thickness of the wing is less than 8 percent of the chord the profile the lower curve at R = 105. If it is assumed that the point of breakdown is uncertain drag is less than that of the equivalent plate. This includes nearly all wings used in between the above limits, the drag of the thin flat plate would be expected to follow Britain before some two or three years ago. the lower curve below R = 105 and the upper curve above R = 5 X 106, but it might Up to a thickness of 12.5 percent the profile drag is within some 10 percent of be anywhere within the shaded area when R lies between these two values. If the that of the equivalent plate. This includes the great majority of wings used in Britain wind tunnel experiment were made on the plate between these two limits of R, the at the present time. drag would be expected to vary erratically from tunnel to tunnel according to the Thicker wings, up to 15 percent or 20 percent, apparently have higher profile turbulence of the air in the tunnel. drag coefficients—between, say, 15 percent and 50 percent greater than the equiva- Our precise knowledge of the drags of very good streamline bodies other than lent plate, according to shape and place of test. the flat plate is confined at present almost entirely to the results of wind tunnel Turn now to Fig. 2. The bulk of the available information here results from tests on solids of revolution, such as bare airship hulls, and on isolated wings. The extensive trials upon two models of bare airship hulls which were tested in a number former when moving axially exert no lift, and consequently their drag, if they con- of different wind tunnels in the various aerodynamic laboratories of the world, form to my definition of “streamline,” should be entirely due to skin friction; the with the primary object of comparing the tunnels. The result was to show that latter, when they exert lift should, if streamline, experience induced drag as well as wind tunnel experiments on very low drag models, such as these, are uncertain and skin friction drag. My first concern, when starting the investigation upon which depend to a great extent on the local conditions, such as turbulence, etc., in the this lecture is based, was to determine to what extent the drag of the better known tunnel. Speaking broadly, it may be said that the various wind tunnels gave experi- streamline solids of revolution and the profile drag (total drag minus induced drag) mental results for these models lying anywhere within the shaded area; you can take of the better known wings, could be estimated from the known skin friction on a your choice as to which you are prepared to believe. There are clear signs, however, flat plate. To do this I have expressed the profile drags of certain wing sections and that as the Reynolds’ number rises towards the extreme limit of atmospheric tun- the drags of certain streamline solids of revolution in the form nels, the uncertainty becomes less and the tunnels tend to agree in placing the coefficient slightly below that of the equivalent plate. This tendency is somewhat faintly 2 kF = (drag)/ρV E reflected in the shape of the shaded area. The black circular dots relate to individual experimental results for these same where E is the total exposed area. These coefficients are plotted in Figs. 1 and two models in the American variable density wind tunnel. These points, which are 2, where they can be directly compared with the curves showing similar coefficients carried to much higher Reynolds’ numbers than were obtainable in atmospheric for a thin flat plate, as explained above. The Reynolds’ number for these wings and pressure wind tunnels, fit the upper or turbulent boundary layer curve with remark- streamline solids is of the form able accuracy, even in the region of the lower Reynolds’ numbers, where the major- R = ρVl/µ ity of the atmospheric pressure tunnels give lower coefficients. The explanation of this is probably that the variable density tunnel employs a fine honeycomb, very where l is the maximum dimension of the body parallel to the wind—the close in front of the model, which should result, according to Burgers’ experiments, “length” of the streamline solid and the “chord” of the wing. in an exceptionally early break up of the boundary layer from laminar to turbulent Comparison in Figs. 1 and 2 between the experimental points and curves for flow. Thus under conditions in which, for a flat plate, the boundary layer would be the solid shapes and the curves for the flat plate at the same value of R, amounts turbulent over the majority of the plate, the drag of two different airship models is to comparison between the drags of the solid shapes and of a thin flat plate of the found to be closely equal over a very large range of Reynolds’ numbers to that of same total exposed area, having a length parallel to the wind equal to the maximum the equivalent flat plate with turbulent layer. length in this direction of the solid shape. I shall in future describe such a thin flat The thin continuous lines near the bottom of the figure, in the neighborhood of plate as the “equivalent plate.” Figs. 1 and 2 are, I think, striking and significant. In Fig. 1 the points clearly R = 106, tend to cluster around the upper curve, appropriate to the turbulent boundary layer, relate to an experiment carried out on a different model in the experimental 138 Chapter 3: The Design Revolution Document 3-7 139

naval tank of the National Physical Laboratory. Here the fluid—water—was very the majority of instances the layer will be mainly turbulent before the Reynolds’ still and the model was pushed from behind instead of being supported by wires number appropriate to aeroplane bodies in flight—say, between 2 and 5 times attached at various points on its surface. External disturbances were thus reduced 107—is reached. to a minimum. This treatment is seen to result in the coefficient falling very close For the purposes of the present paper it is sufficient to note that the upper or to the lower curve appropriate to a flat plate with laminar layer. The experimental turbulent layer curve appears to give an upper limit to the drag coefficient of good curves, however, begin suddenly to rise at Reynolds’ numbers slightly greater than streamline shapes, whether they be in the form of wings or solids of revolution, 106, in exactly the manner which would be expected if the layer at the rear part of provided that the fineness ratios are not less than 8 in the case of wings and 4 in the the model were beginning to become turbulent at the higher Reynolds’ numbers. case of solids of revolution. That is to say, the curves rise approximately along the theoretical transition curves, The information displayed in Figs. 1 and 2 relates to simple shapes such as which here have a slightly different shape from those in Fig. 1, owing to the tapered isolated wings and solids of revolution. There is as yet little information available tails of the models. This same model was tested with the same method of support in relating to good streamline shapes in combinations, such as the wings and bodies of one of the National Physical Laboratory 7 ft. wind tunnels, and produced a curve aeroplanes, or to the distortion and interference to which a shape can be subjected almost coincident with the lower of the skin friction curves in this figure. It was without becoming unstremline . Such information as is available suggests that com- not, however, practicable in the wind tunnel to reach a Reynolds’ number much binations, or minor distortions, can be made without appreciable increase of drag, above 106 and no signs of the rise in coefficient was observed. These experiments, provided we know how to make them. It is therefore reasonable to suppose, after both in air and water, were repeated with a thin string loop slipped on to the nose examining Figs. 1 and 2, that complex bodies such as aeroplanes could, given suf- of the model, to form a slight obstruction which, from previous experiments of a ficient knowledge and structural ability, be made to have a drag no greater than the similar nature, would be expected to cause the boundary layer behind it to become sum of the induced drag and the drag of the equivalent flat plate. turbulent. With this addition the drag coefficients both in air and water increased If the drag of the streamline aeroplane is simply the sum of the induced and skin some fivefold and approximated closely to that for the equivalent flat plate with friction drags, the power required to tow it would be obtained by multiplying these turbulent boundary layer. Thus, in circumstances where the boundary layer, from drags by the forward velocity V. Let I and F be the powers required to overcome analogy with the flat plate, may be expected to remain laminar to a relatively high induced and skin friction drags respectively. If the streamline aeroplane is towed by Reynolds’ number, the drag of the model is very closely equal to that of the equiva- an airscrew of efficiency η, effectively isolated from it so that there is no interfer- lent flat plate with laminar boundary, and the rise in coefficient, when it does occur, ence, then the b.h.p. required to drive the screw will be takes place in exactly the manner which would be expected on a flat plate with tapered rear. When, however, a slight change is made to the surface, which would (I + F)/η be expected to break up the boundary layer over the greater part of the surface, the drag approximates closely to that of the equivalent flat plate with turbulent bound- How are we to regard interference between screw and body, when such exists? ary layer. Interference will alter torque and thrust reactions between screw and engine. The I have examined the drags of other streamline solids of revolution in the same change in torque may call for a redesign of the screw to make it balance engine way as above, and I find that so long as the fineness ratio is not less than 4 the coef- torque at suitable engine revolutions. The change of thrust is of no interest, except ficients all fall between the upper and lower curves of Fig. 2. in relation to the size of thrust bearing required, for it may be largely neutralized by Reviewing the evidence relating to streamline solids of revolution, it seems dif- change of body drag. Neither change is of direct interest from the present point of ficult to resist the rather surprising conclusion that their drag can be estimated with view, which is to examine any possible changes in b.h.p. due to interference, assum- considerable accuracy from the drag of what I have called the equivalent flat plate. ing always that the screw is properly designed for the engine. With this assumption it is If the experiments recorded in Fig. 2 had been conducted upon thin flat plates of not difficult to show that, to a first approximation, the b.h.p. required to propel the the same superficial area as the models and of the same length in the direction of streamline aeroplane should not be seriously influenced by interference, provided motion as the overall length of the models, they cold not have given results more in that the interference does not cause the flow to cease being streamline. conformity with the known behavior of the boundary layer on a flat plate, than the We arrive at this conclusion by considering the disturbances left in the air results actually recorded. behind the streamline aeroplane. They may be divided into:— We do not yet know to what extent it may be possible to increase Reynolds’ 1. The airscrew slipstream. numbers whilst retaining a laminar boundary layer, but it seems probably that in 2. The induced vortices. 140 Chapter 3: The Design Revolution Document 3-7 141

3. The small scale turbulence, and possibly temperature rise, due to skin fric- for the increased velocity over surfaces exposed to the slipstream and I to allow for tion on the exposed parts of the airscrew blades and on the aeroplane. alterations in lift distribution across the wing span due to the slipstream. In the esti- Since, by hypothesis, the flow around the streamline aeroplane is substantially mates which follow the first approximation is sufficient for my purpose and I have that of an inviscid fluid, except in the very thin boundary layer where the skin fric- not proceeded beyond it. tion is applied, the whole energy expended—the b.h.p.—is included in the above When the streamline aeroplane is climbing, the above expression for brake three items. horse-power must include another term C, equal to the product of weight and rate The energy in the slipstream is mainly due to the axial velocity of the stream; of climb. Thus it is dependent upon the effective diameter of the stream and the net thrust of the b.h.p. = (F + I + C)η screw and body behind it. If, as we may reasonably suppose, these are unaltered by the interference, item (I) will remain substantially unaltered. It remains to give numerical expression to the symbols I, F, C and η.

The energy I in the induced vortices will not be altered by interference, unless The estimation of I presents no difficulty. A simple formula I = 2.80 ϖ/σVm the interference appreciably alters the lift distribution across the wing span. If con- horse-power per thousand pounds weight with span loading ϖ in pounds per sq. ft.

sidered necessary, the alteration could be estimated and I taken to apply to the and Vm measured in hundred-mile-per-hour units has already been given. This for- induced power associated with the actual lift distribution in the presence of inter- mula applies strictly to a biplane or normal gap and aspect ratio and of rectangular ference. plan form, but is easily modified (see footnote on p. 4) to apply to a monoplane or The energy expended on skin friction on the airscrew blades will not be seri- biplane of any desired gap or aspect ratio. ously affected by any reasonable interference. The skin friction power F depends upon the skin friction coefficient 2 The energy F expended on skin friction on the aeroplane will be affected only kF = (skin friction drag)/ρV E

on those surfaces which are exposed to the slipstream. This effect will always be and before F is estimated some value for kF must be adopted. The question

small; at high speeds, where F is an important item, the slipstream velocity is low, arises as to what value is to be given to kF. If it is to be taken from the value for the whilst when climbing F is but a small part of the power expenditure. To a first order, equivalent flat plate, i.e., from the upper curve of Figs. 1 or 2, its value will depend therefore, this item in the power account will also be unaffected by interference, but on the Reynolds’ number, and this raises the further question as to how the Reyn- if higher accuracy is required F must be taken to relate to the estimated skin friction olds’ number for the aeroplane is to be estimated. The Reynolds’ numbers of full with allowance for increased velocity over surfaces exposed to the slipstream. scale aeroplane wings, based on the chord as a measure of size lie between 4 X 106 7 I do not maintain that the above arguments are precise, but if they are examined and 10 and the corresponding value of kF for a thin plate (see Fig. I) lies between carefully, with due regard to the quantities involved in practicable aircraft, I believe 0.0020 and 0.0017. The Reynolds’ number of the bodies of full scale aeroplanes, that they will be found sufficient. based on their length as a measure of size, lie between 2 X 107 and 5 X 107, and

We are now in a position to state the main conclusion of the lecture, which is the corresponding value of kF (see Fig. 2) lies between 0.00155 and 0.00135. The as follows:— logical proceeding would be to estimate the wing skin friction drag from Reynolds’ The brake horse-power required to propel the streamline aeroplane horizontally number appropriate to the body. But for simplicity and so as to be on the safe side,

may be estimated as I propose, for my present purpose, to take an over-all figure for kF of 0.0020. This (F + I)/η should leave us a little in hand to allow us to use moderately thick wing sections— say, thickness up to 12 percent of the chord—in our ideally streamline aeroplane. where η is the efficiency to be expected from an isolated screw performing the Using this figure we have service required. F is the power expended in skin friction, which may for the present be taken to F = 0.002 ρV3E foot pounds (using any consistent units). 3 be that required to propel the equivalent flat plate. = (27 σVm E)/W horse-power per 1000 lbs. weight.

I is the induced power. where Vm, is in hundreds of miles per hour units, E is the exposed surface in square feet, and W is the weight in pounds. As a first approximation F and I may be estimated for the aeroplane isolated Finally, the climb power C, per 1,000 lbs. weight, is easily shown to be given by

from the screw. As a second approximation F can be altered to allow approximately C = 0.030 VC where VC is the rate of climb in feet per minute. 142 Chapter 3: The Design Revolution Document 3-7 143

These three simple expressions for I, F and C, allow an approximate and probably conservative estimate to be made of the power that would be required by an aeroplane in any given circumstance, provided that the flow around it were everywhere streamline and no unnecessary eddies were generated. The difference between the actual power expended and the power so estimated must be regarded as being expended in the generation of unnecessary eddies which might be avoided by more careful attention to external form. I have thought it of interest to prepare a few curves showing the performance, in level speed and climb, of a streamline biplane in air of standard density near ground level. For this purpose I have taken E = 3.2 X S where S is the conventional wing 3 area, so that F = 86 σVm /ω where ω is the wing loading in pounds per sq. ft. I arrived at this value 3.2 by computing E roughly for a number of well-known aeroplanes and finding that E/S ranged between 3.0 and 3.5 with 3.2 as an average value. In drawing the curves I have taken σ to be unity and η, the screw efficiency, to be 75 percent for level flight and 70 percent climbing. Fig. 3 shows four curves giving b.h.p. per 1,000 lbs. weight, for level flight at various speeds for, the following conditions:—

wing loading (ω) 10 and 15 lbs. per sq. ft. span loading (ϖ) 2 and 2 ½ lbs. per sq. ft.

The individual points, with the names of well-known aeroplanes written against them, refer to the b.h.p. at top speed as given in the 1927 edition of Jane’s “All the World’s Aircraft.” The small circles relate to aeroplanes for which ω is approximately 10 and the squares to those for which ω is in the neighborhood of 15. The vertical distance between the points and the appropriate curve represents the power being expended on what, from the present point of view, much be described as unnecessary turbulence. Horizontal distances from the curves represent the increased velocity which might presumably be obtained for the same power expenditure in the absence of unnecessary turbulence. Apparently large commercial aeroplanes, such as the Argosy, the W.10 and the Hercules, would, were they ideally streamline, either fly at the present top speed for one-third the present power, or alternatively travel some sixty miles per hour faster for the same power. Two-thirds of the power used by these aeroplanes is being expended on the generation of turbulence which, from the aerodynamic viewpoint, is unnecessary. Fig. 4 is constructed on the same lines as Fig. 3 except that the abscissae represent rate of climb and the three continuous lines show the expenditure of horsepower required for a streamline aeroplane, divided up into the three parts: weight- lifting power, induced power and skin friction power. Here the span loading is 2.2 lbs. per sq. ft., which is a good average figure for modern aeroplanes, whilst the wing loading is taken as 10. The airscrew efficiency is taken to be 70 percent. 144 Chapter 3: The Design Revolution Document 3-7 145

Though improvements in power consumption for climbing are not so great ficulty was no doubt as difficult for them as for us, but it has ended in the complete as for top speed, they are still appreciable. The very high ratio between the power triumph of the external form. We can only hope that it will not take us so long to apparently expended in turbulence in some cases and the power lost in skin friction, reach this point as, if we are to believe the comparative anatomists, it took them. suggests very high turbulence losses, due probably to severe interference between the slipstream, the wing and the body in these circumstances. Possibly the assumed DISCUSSION airscrew efficiency of 70 percent is too high for climbing conditions on modern The PRESIDENT: Professor Melvill Jones had drawn the attention of the commercial machines, and this may account for some of the heavy losses indicated Aeronautical Research Committee a year or so ago to the matters dealt with in the in the diagram. paper; as a result, experiments were carried out, which had in substance proved his contentions. He had set up a valuable standard, and when one considered how far Recapitulation present-day aircraft, both military and civil, were below that new standard, one The lecture consists of a search for a simple and reasonably sound way of esti- could hardly cavil about its supposed inaccuracies. He had stated in the paper that mating the power which would be required to drive a perfectly streamline aero- his conclusions were based upon leaky theoretical arguments, but one felt that plane. The formula finally reached is very simple and easily visualized; the difficulty when the resistances of present-day aircraft had been reduced by about 50 percent lies in defending the approximations by which it is reached. it would be time enough to consider those leaky theoretical arguments with a view There are two main steps in the argument: the first is that the power required to to their revision. tow the ideally streamline aeroplane should be merely the sum of the induced and Major F. M. GREEN: He thanked Professor Melvill Jones for having set up not skin friction powers. This idea suggests the use of a drag coefficient, based on the merely an ideal, but an ideal which could be expressed in figures. It meant much total exposed area, to estimate that part of the power which is not absorbed in the more to a designer to be able to say that his aeroplane represented a definite advance easily calculable induced drag and in climbing. towards a definite ideal, than to be able to say merely that it was a little better than A comparison of such drag coefficients for a series of simple streamline shapes, its predecessor, because in the latter instance he did not know how much further with the skin friction coefficient of a thin flat plate, with turbulent flow in the he had to go. Designers were now aware that a great deal of progress must be made, boundary layer, suggests that the latter provided a convenient and safe estimate for and the paper would stimulate them to make better aeroplanes. The influence of the drag coefficient of any good streamline body. figures was very great, and designers would be able to express the efficiency of their Finally, a figure of 0.002 is shown to give a conservative estimate of the coef- aeroplenes as a percentage of Professor Melvill Jones’ ideal, instead of saying merely ficient for wings and bodies at the Reynolds’ numbers of full scale flight. that a particular aeroplane was fairly well streamlined or was fairly efficient. With The second step in the argument is that in the streamline aeroplane, interfer- regard to the overall figure of 0.002 for the skin friction drag of planes, he said ence between the airscrew and the aeroplane, though it may affect the design of that was half what was at one time called the minimum profile drag (0.004). The the screw suitable to a given engine, should not greatly affect the over-all power figure would vary, however, with different angles, and with greater angles the figure absorbed; hence, when estimating the brake horse-power required for the streamline might be greater. With regard to turbulence, that in the air occurred on a very much aeroplane, the screw efficiency may be assumed to be that of an isolated screw. greater scale than in a wind tunnel, and he asked whether the results obtained with These two conclusions, if accepted, enable the power required by the streamline an aeroplane on a day when there was considerable turbulence would be appreciably aeroplane to be very easily calculated. It is submitted that this should always be done different from those obtained on a calm day. in order to keep before the designer and other people concerned the price that is Mr. R. McKINNON WOOD, referring to the use of the wind tunnel, said being paid for defective streamlining. that for many years it had been thought that the ideal condition of test was that It is not suggested that it is easy to design a streamline aeroplane which will be the turbulence of the air in the tunnel should be the minimum attainable, but in also a practicable machine, but the immense saving in power, and therefore in fuel the last year or so it had dawned on some investigators—it had dawned on him a consumption, which would apparently follow such a step, forces me to the belief few months ago—that possibly the wind tunnel which was regarded as the worst that design will evolve steadily in this direction and that the ultimate aeroplane will was really the best; because, if there were sufficient turbulence in the air the results be as well streamlined on the whole of its external surfaces as, say, the bottom of a obtained would be as indicated by the upper curve exhibited by Professor Melvill racing yacht or the externals of an albatross. I am fortified in this belief by survey- Jones, and if one extrapolated along such a curve one should make a fairly good ing the animal kingdom. Those birds and fishes which depend on speed for their prediction of the full-scale value. If the results lay below that curve—somewhere in existence have long since solved the problem. The compromise with structural dif- the indefinite area—one could not say in the least what the full-scale value might 146 Chapter 3: The Design Revolution Document 3-7 147

be. He hoped that the paper would be of great service by focusing the attention of large amount of power frittered away in driving modern aeroplanes, and the paper designers of aircraft more on aerodynamic considerations. He would not reproach should prove a great stimulus to all workers in aeronautics. designers for giving, as he considered they did, too much attention to weight, to Mr. C. C. WALKER: Those concerned with aviation were particularly inter- mechanical, structural and other aspects of their work and too little to the aero- ested to know what chances there were of recovering some of the lost power and dynamic aspect. The reason could be found very simply; one could determine the how much of the loss was inevitable. When Professor Jones had first published weight of a machine easily and precisely, but it was more difficult to determine drag some of his results, he (Mr. Walker) had made calculations in respect of a number and to determine what improvements were effected by small alterations. A question of aeroplanes, taking, instead of approximations, the actual “wetted” surface of each which was continually before those engaged in full-scale test was what improvement aeroplane and the available h.p. (after multiplying the b.h.p. by the propeller effi- was effected by replacing, say, a radiator by a different radiator. The gain in speed ciency) and had found that an ordinary commercial aeroplane did attain from 60 effected thereby was probably half a mile or one mile per hour, but the man who to 67 percent of the streamline speed. Both the best and the worst examples among was making the tests knew that variations in engine power, up and down currents, his own company’s machines happened to be monoplanes. In a little racing machine and other factors might affect speed to a greater extent than would the changing of which did 197 miles per hour with 133 h.p., a little over 80 percent of the stream- the radiator. He believed that this work of Professor Melvill Jones, in that is showed line speed was attained. That meant that what had to be recovered amounted to how much better a performance we could look forward to, would have the effect of the equivalent of three-quarters of a square foot of flat plate presented normally to focusing the attention of designers more and more on the problem of reducing drag. the wind, and that had to include the undercarriage, all the bracing, the cooling of Mr. A. FAGE: He would like to mention two series of experiments with which the engine, the cockpit, and the effects of the slipstream on the body and wings. he had been connected, and which had a direct bearing on Professor Jones’ paper. It The machine must have an undercarriage, and at present it must have some form was well known that on a large circular cylinder there was a certain range of Reyn- of cockpit; cooling resistance might be reduced somewhat, but there seemed to be olds’ number (VD/v) over which the character of the flow underwent a marked very little margin for recovery in a machine of that sort. In the case of a commercial change, and which was accompanied by a large drop in the drag coefficient. He aeroplane, which must have all sorts of other excrescences, it was difficult to fore- had had occasion to explore the boundary layers for these flows and to determine see how far one could go towards eliminating the resistance. Such a machine had the breakaway points. Denoting the circular distance from the stagnation point to a radial engine sometimes on the front of the body, the cabin had to be ventilated the breakaway point by “l,” and the velocity just outside the boundary layer at the by means of structures like ships’ ventilators, sometimes there was a starting engine breakaway point by “V,” he had found that the range of Reynolds’ number (Vl/v), mounted outside the fuselage, and it did not matter very much what one did in over which the marked change of flow occurred was from 1.1 X 105 to 4.4 X 105. the way of fairing after that. It would be interesting to know what Professor Jones This was exceedingly interesting, because, as mentioned by Professor Jones, Profes- thought about the amount that should be saved. (Laughter.) Referring to the prob- sor Burgers had found that the critical range of (Vl/v) for a flat plate (l is the distance lem of induced drag h.p., he said that Lanchester, in 1914, had shown how to work from the leading edge) over which the flow along a flat plate could break down was it out, and from his book, which was published in 1908, one found that he had put about the same order, namely, 105 to 5.0 X 105. There was, therefore, from this point it forward to the Physical Society in 1894. At that time he had called it “aerody- of view, a resemblance between the flows in the boundary layers along a flat plate namic drag,” but it appeared to be the same as “induced drag.” He (Mr. Walker) did and around a cylinder, and we should expect intermediate bodies such as airship not know whether or not Prandtl had worked it out earlier than that, but the point forms and aerofoils to exhibit similar tendencies; Professor Jones had shown that was that if “aerodynamic drag” were the same as “induced drag,” and nothing more they did. Another point of interest arose from some experiments made to measure than a change of name was involved, it was rather a pity that Lanchester should be the minimum drags (at zero incidence) of a family of symmetrical Joukowski sec- deprived of the credit due to him. tions of different thickness. If one plotted drag coefficient against thickness one was (Communicated): There is an aspect of Professor Jones’ criterion of performance able to predict the drag coefficient for a Joukowski aerofoil of zero thickness (i.e., which is not free from objection. Presumably, by encasing the various necessary a flat plate). The values of Reynolds’ number at which the predictions were made excrescences in large streamline boxes, or by so increasing the fuselage in size as to were in the critical region, and as would be expected the predicted values of drag embrace what otherwise would not be embraced, credit could be taken for the extra coefficient lay between those for laminar and turbulent flows along a flat plate; fur- surface, and a high streamline efficiency shown. ther, these predicted values agreed fairly closely with Geber’s values for a flat plate It could then quite easily happen that of two aeroplanes doing the same job, the at the same Reynolds’ number. These results gave additional support to Professor slower would show a better figure of merit on this basis than the faster. Jones’ conclusions. A great merit of the paper was that attention was focused on the In one of Professor Jones’ earlier papers, he placed on the credit side the area of 148 Chapter 3: The Design Revolution Document 3-7 149 the floats of a seaplane—as no credit can be taken for the wheels of the correspond- tests could almost be applied directly to full-scale without correction. Further, the ing land machine—the situation mentioned above may be said to have already parts were of fairly fine form and the minimum drag coefficients were low. In one arisen. machine the model results indicated a higher maximum lift than was expected, but The fact remains that the great virtue of this method is its absolute nature. the performance of the machine was found to agree remarkably well with all the For purely comparative purposes, it is perhaps preferable to give the highest model test results. figure of merit to that aeroplane which carries a given paying load per h.p. fastest in Reverting to the skin friction curves given by Professor Jones, it would be noticed relation to its landing speed or to use some similar basis. The “streamline efficiency” that the phenomenon of double flow and the transitional region was similar to flow will then show to what extent this machine falls short of the ideal, but to aim directly in pipes, but in the latter it was usual to make in the expression log Vl/v equal to and only at this in design would not necessarily produce the best aeroplane. the diameter of the pipe and not to a down wind dimension. Now it was implied Mr. W. L. COWLEY: It would appear from his remarks that the wind tunnels by Professor Jones that surface area at the rear of the body was affected differently had given us fictitious results in the past, but he would like to hear the lecturer from that near the front on account of the accumulated disturbance produced in the explain the fact that, in many cases, excellent agreement had been obtained between boundary layer by the action of the surface coming before; and the average drag per model results and full-scale. As far as his personal knowledge went, designers had unit area of the surface of a good streamline form depends only upon the length of the greatest faith in wind tunnel work. The host of results upon various wing, strut, the body when the speed, density and viscosity were constant. Now the resistance of body and float forms tested in the past as well as on complete models, airscrews, a parallel pipe was all skin friction and its drag coefficient appeared to be of the same controls, stability, etc., were made use of daily and no marked discrepancies with order as that for the flat plate. One would expect, therefore, similar conditions to full-scale, except in occasional cases such as R.A.F. 19, had been reported. It would prevail. Thus, if for the flat plate, the boundary layer became turbulent at a distance be strange if these agreements were purely accidental and still more so if they were l from the leading edge, then it would be expected that turbulence commenced in a but imaginary, and that this fact had escaped the great army of workers that had pipe at the same distance from the entry. In other words, turbulence in a pipe com- been engaged upon the subject. menced at a certain distance from the end irrespective of the diameter, so that, for Just before attending the lecture he calculated the drag coefficients of an aero- two pipers a and b using the same fluid, the distances and speeds were related by foil, a body and a gloat that were tested some time ago in connection with a certain

machine. The points fell almost exactly upon the turbulent boundary layer curve Vala = Vblb. given by Professor Jones. Now Professor Jones assumed that the drag curves for the airship forms ultimately merged into this turbulent boundary layer curve. If it The dimensional theory, however, demanded that, if the conditions were simi- be assumed that the drag curves for the parts mentioned should also follow this lar, a breakdown in flow would occur at corresponding points when curve, the drag coefficient for the full-scale machine should have been approxi-

mately 0.0014 instead of about 0.0022 as the model scale effect curves indicated. VaDa = VbDb The full-scale result appeared to be more consistent with the latter than with the former result. It may be argued that the resistance measured were only partially skin where D is diameter. In other words, the diameter is all important. It followed, friction. If that was so one would expect the difference between 0.0022 and 0.0014 therefore, that the main body of the fluid affected the resistance as well as the surface to be form resistance and the curves of drag coefficients of the models against log Vl/ fluid. v to follow a transition curve rising up rapidly to a turbulent boundary layer curve The apparent agreement between the experimental pressure distribution round parallel to the one given, but having the ordinates greater by an amount equal to a body and the pressure distribution calculated from the inviscid fluid theory gave this difference. In the tunnel tests mentioned, the variation of drag coefficient with no support to the contention that the integral effect on drag of the flow in the main log Vl/v however, was very slight. During recent years tests have been made at the part of the fluid was negligible. After all the integral of the discrepancies between National Physical Laboratory upon several models of complete machines, ranging the pressure distributions, when resolved, was equal to the whole of the drag. One from single-seater single-engined machines to large three-engined machines. Parts could only conclude that the inviscid fluid theory, in its present form, might be were tested alone and in combination and the whole work has been carried out in good enough for calculating certain effects, but was useless for any work involving direct touch with the designers of the machines concerned. In no case has there profile drag. been the slightest cause to suspect the results. It should be noted that the scale effect In conclusion, he said that although he regarded the lecture as very interesting was considered so small that the results at the highest values of Vl/v in the tunnel and instructive, he thought it was a little dangerous to act upon any deduction made 150 Chapter 3: The Design Revolution Document 3-7 151 from it at the present stage of our knowledge. It did not justify as yet any loss of faith Jones’ lower curve was based on measurements of flow near flat planes and not on in wind tunnel work or any change in wind tunnel technique, although he agreed overall measurements of the skin friction on a balance. that further research on the lines given by the lecturer might be necessary. Mr. CAPON: On a previous occasion when Professor Jones had lectured to the Mr. PYE: Referring to Professor Jones’ curves showing the b.h.p. per 1,000 lbs. Society, he had uttered the dictum that no research could be considered to be com- weight in respect of certain machines for level flight at various speeds, he suggested peted until it had been reduced to a rule of thumb. One might add that no research that one or two spots representing the performance of machines which competed in should be begun until there was a definite formula to work to, and the great value the Schneider Trophy contest might be added, because the drag had been very mate- of Professor Jones’ work was that he did provide such a formula. It had been known rially reduced in the case of those machines. If such information were given, one for a long time that great savings in drag could be effected; probably the reason why would be able to make extraordinarily interesting comparisons, because the drag of so little had been done was that there were a great many small contributions, and those machines must be very much nearer to Professor Jones’ theoretical limit than nobody felt really convinced about any of them. For example, it had been known it was for any of the machines he had shown. for several years that a saving in drag on models could be effected by cowling the Dr. DOUGLAS: No doubt in cases where the full-scale flow was turbulent cylinders of radial engines. Why were radial engines not cowled? Probably because increase of turbulence produced a similar effect to that obtained by increasing Reyn- designers quite reasonably thought that there might be a scale effect on drag or cool- olds’ number, and a better idea of full-scale value would be obtained if models were ing, or both. Again, we really did not know how much the sum of these contribu- tested in a very turbulent tunnel, as suggested by Mr. McKinnon Wood. It was quite tions would amount to. Professor Jones had given a figure of merit to indicate what possible, however, that on objects, such as struts, the full-scale flow might actually could be expected, and although there might be some quibbles as to what exactly be laminar. that figure amounted to—questions such as airscrew interference, and so on, were Professor Jones had shown that existing aeroplanes have very much more drag involved—it was nevertheless a sufficiently definite figure. Professor Jones was to be than they need have. Machines were defective both in form and in surface. The congratulated, not only upon having produced it, but on having met so successfully, latter defect was particularly noticeable on the Schneider Trophy winner displayed as it appeared, the objections raised by the scientists based on the uncertainty as to in London, and he had been shocked to notice the rivet heads and other excres- the skin friction coefficient to be adopted. His investigations in this direction had cences on a machine designed to have the lowest possible drag. The components apparently led to results of great interest in other connections. of a machine of this type have drags little greater than that given by thin plates of Mr. Scott HALL: It seemed to him that Professor Jones’ assumptions with similar surface area, and when this standard had been reached it must pay to keep regard to the working of the screw in close proximity to the body were of a very the surface clean. He hoped that Professor Jones’ remarks would lead to a better grave nature, for in effect he had assumed that the loss due to increased drag of the appreciation of the need for reducing drag. body in the slipstream behind the screw was balanced entirely by the extra efficiency Mr. IRVING: With regard to the point raised by Dr. Douglas, Professor Jones, with which that screw worked, due to the body behind it. He would have liked to in arriving at his overall figure for the full-scale drag coefficient, had assumed that have heard more from Professor Jones on that matter. With regard to the curves one traveled along the upper curve. It was just conceivable, however, that with cer- showing the performance of various aircraft, he said that the point relating to the tain shapes one might possibly keep to the lower curve. The departures from the “Spirit of St. Louis” was situated a considerable distance from those relating to other lower curve had in most cases occurred in wind tunnel tests in which the general machines. The “Spirit of St. Louis” was the only machine dealt with which had no flow was turbulent, although in one case, in which tests were made by towing a open cockpit, nor openings of any kind, and the fact seemed to indicate that the model in still water, there was also a departure from the lower curve in the direction openings on a streamline body were of very serious practical importance, more so of the upper curve. Mr. Irving also referred to some tests carried out at the National even than had been hitherto supposed. Physical Laboratory in 1919, in which Mr. Ower and he had measured the skin Mr. BRAMSON: He could not resist the temptation to point out one analogy; friction of flat plates of various thicknesses on the balance in the wind tunnel, and in all probability Professor Jones’ treatment of the subject of the streamline aero- extrapolated from different thicknesses down to no thickness. The values of the skin plane would be regarded as holding the same position in relation to aerodynamics friction coefficient obtained for different sizes of plate at different wind speeds were as, for instance, the Carnot cycle held in relation to thermo-dynamics. Professor plotted against VL; for the lower values of VL the points came actually on the lower Jones had put forward an ideal beyond which, without the discovery of new prin- curve, but as VL increased there was a gradual departure, and at the higher values ciples, one could not hope to get, and he had given designers a standard by which of VL the curve crossed the region between Professor Jones’ two curves and almost the qualities of aeroplanes could be judged, a standard which has hitherto been touched the upper curve. He mentioned this fact because he understood Professor lacking. He asked Professor Jones if he could give, there and then, an actual numeri- 152 Chapter 3: The Design Revolution Document 3-7 153 cal example of what an aeroplane of reasonable weight, horse-power, and so forth, tions being broken at their points of juncture with the body. Such a condition must must do to achieve the ideal. With regard to the curious fact mentioned by one of cause mutilation of the air flow at these points and give rise to additional drag the speakers, that in the course of experiments he had found that the lift coefficient (now termed interference). In order that these disadvantages can be obviated, the began to drop at much lower values at low values of Reynolds’ number than at high writer has designed several types of landing gears where the structure is rigid and the values, Mr. Bramson asked if it were not possible that that was due to the departure shock-absorbing mechanism lies within or adjacent to the wheel. The wheel with from the boundary layer occurring sooner at low speeds, due to the relatively greater internal springing consists of an oil dashpot shock absorber, compression rubber effect of the “adverse pressure gradient” near the trailing edge, than at high speeds. springing and wheel brakes. This design of wheel has been in existence for two With regard to circulation, he said he had heard it stated that the “equivalent circu- years, but it has only recently been favorably reviewed by wheel manufacturers. The lation” round an aerofoil would never start in the ideal fluid. He asked if that were Curtiss Company, of America, have produced a similar type of wheel, and they con- so, and whether the idea of a circulation around an aerofoil was at all a sensible one. sider this undercarriage an attractive proposition. The shock-absorbing capacity of It was difficult to see how, if there were anything which could properly be called these wheels is better than that usually obtained with an oleo leg, because there are circulation, it could go round a sharp trailing edge, for instance, in any intelligible no limitations due to angular movements of the structural members. Conservative manner. detail estimates show that the undercarriage drag can be reduced by 25 percent and George H. DOWTY (contributed): Following Mr. North’s recent lecture, it is the weight by 15 percent. possible to compare the ideal aeroplane with present day standards and appreciate G. T. R. HILL (contributed): This paper sets up for the first time a standard by the vast improvements which have yet to be made. The lecturer stated that prog- which the performance of any aeroplane may be judged. Considering how far all ress towards the ideal streamline aeroplane has been slow and points out that the designers are below that ideal performance, or “Jones performance,” as it should now tendency to belittle minor gains is a wrong policy. The design of racing machines be called, serious objection cannot yet be brought against the standard on account has produced a very clean type of aircraft, and here the designer is prepared to sacri- of its being built up with the aid of what Professor Jones calls “a leaky theoretical fice practically everything in producing an ideal aerodynamic form. When speed is argument.” When the present drag of our military and commercial aeroplanes has not the ruling factor, the question of maximum aerodynamic efficiency often takes been halved, it will be time to set the standard more securely on its base with the aid second place to other considerations and justification is claimed for this because the of aerodynamic knowledge which will then be available. increase in performance is small. If the existing gap between present designs and the There appears no doubt that the rate of progress towards the streamline aero- streamline aeroplane is to be bridged then improvements, however small, cannot plane is greater than that achieved by Nature in bygone ages in the design of birds be neglected. One of the chief items of parasitic drag, and one to which Professor and fishes, yet any elation which may be felt over this will be considerably moder- Melvill Jones has drawn attention in R. & M. No. 1115, is the undercarriage. The ated by turning back some eighteen years to 1911, to the remarkable designs of writer has been engaged on work in this direction for some considerable time, and M. Nieuport. There is a good description of these little aeroplanes in the old Aero during the last three or four years has confined his attention to methods for reduc- of June, 1911; one of them which flew in the Gordon-Bennett Race of that year ing the drag of this unit. Investigations have led to the same conclusions as those was fitted with an engine of only 28 h.p. nominally and with a wing area of about given by Mr. North, that the retractable undercarriage is not desirable because of the 170 square feet, it is reported to have attained a speed of 75 m.p.h. Looking at the complicated retracting mechanism and the extra weight involved. Mr. North sug- photograph of the aeroplane it might be asked how could its resistance be brought gested the use of mechanical launching gear, and from his remarks, it is understood, down to the modern standard? It could be said at once that streamline wires could a form of ski landing gear. This form of undercarriage is open to strong objections be fitted instead of round cables, and having said this, nearly all has been said. because it limits the operation of aircraft to stations equipped with launching gear Turning to the future, the fact cannot be ignored that as the resistance due to and such machines would be at a great disadvantage in the case of a forced landing. some parts of the aeroplane is reduced, the resistance of those parts left unaltered There is also the question of ground maneuverability and it does not seem reason- calls more and more loudly for attention. For this reason, it is difficult to foresee able to expect any sacrifice in this direction. clearly the future of the air-cooled engine, in which perhaps four-fifths of the resis- There is fortunately some scope for cleaning up existing forms of undercar- tance is due to the turbulent air flow around the cylinders and one-fifth to the skin riages, and it is on this basis that the writer has been working. The shock-absorbing friction which necessarily accompanies the cooling effect required. Already the air- member is usually located in the slipstream and it will be found that this member cooled engine is seen to be at a disadvantage in high speed racing aircraft, and in the gives a resistance at least five times as great as a steel streamline tube carrying the usual course of events, characteristics of the racing craft of today are exhibited by the same load. Further, the existing types of connections are very poor, the strut sec- general purposes craft of tomorrow. How long will that tomorrow be in coming? 154 Chapter 3: The Design Revolution Document 3-7 155

The “Jones ideal performance” stands out as one of the beacons towards which the designer must steer his painful way through the fog of detail work which necessarily surrounds him. S. J. DAVIES and C. M. WHYTE: (contributed): The writers would like to thank Professor Melvill Jones for the excellent way in which he made a highly specialized subject both intelligible and interesting to those who, like themselves, are not conversant with the subject. They were particularly struck by the number of points arising which threw light upon matters which had been obscure to them in an allied subject. It is possible that the following remarks, based on the study of flow through pipes, may in their turn be of interest to aeronautical engineers. One of the most striking points brought out by the lecturer was that the majority of wind tunnel tests lie within the transfer region between laminar and turbulent flow. This fact must add greatly to the difficulty of interpreting the results, and it is remarkable that reliable figures have been obtained at all. For many years the similar transfer region in connection with pipe flow also gave rise to difficulty and led to much confusion of thought. And in fact it is only within the last few years that any certainty has been reached. The curves from flow in pipes are similar to those given in the lecture and the results of the former may help in interpreting the latter. The work of L. Schiller has played a large part in demonstrating that the amount of turbulence in the fluid as it enters the pipe determines the value of the Reynolds’ number at which the transfer takes place. Some of his results, which do not appear lecturer. The value of the Reynolds’ number at transfer, however, remains approxi- to have been published in this country, are shown in Fig. 5. The wide difference mately unaltered. The lecturer’s method of calculating the form and position of the between the values of the Reynolds’ numbers of the figure and those mentioned in transfer curve is interesting, but he does not indicate the grounds for his implied the lecture are due to the fact that in the figure the Reynolds’ number is that formed assumption that the Reynolds’ number, formed by the distance from the leading by the linear dimension of the cross section of the stream, while in the lecture the edge to the breakdown point, is constant during the transfer period. At first sight it overall length of the body is used. The value of F/ρv2 in the figure is arrived at by might be expected to decrease progressively as the transfer curve is traced out in an the method used in the lecture, but it should be noted that it does not include the upward direction. resistance at or near the entrance to the pipe. The various curves were obtained with The lecturer gave the impression that curvature tends to cause the boundary increasing entrant disturbances, and show that the greater the turbulence at the layer to become turbulent sooner. It would be interesting to know if there is any inlet the smaller is the Reynolds’ number at which the transfer takes place. There is, experimental evidence supporting this view. Curvature in pipe flow, provided that it however, a limiting value, and no amount of disturbance will cause the transfer to is slight, appears to be without influence, but, as is shown by some tests about to be take place below R = 1160. The relevant conclusion to be drawn from this is that, published, flow through a curved pipe, of which the radius of curvature is 15 times unless a high degree of turbulence is provided by artificial means, the results will the pipe radius, is considerably more stable than in a straight pipe, and the transfer neither be free from the influence of the unknown state of the fluid at entry, nor does not take place until about four times the Reynolds’ number at which it would be consistent among themselves. With artificial turbulence it is possible to obtain occur in a straight pipe. points on the turbulent line for flows only slightly in excess of R = 1160. Further The necessity for carefully regulated conditions when determining the resis- tests of Schiller show also that a moderate degree of roughness of the pipe does not tance of complicated bodies is well illustrated by tests of a series of very rough pipes influence the transfer. carried out under somewhat indefinite entry conditions. These tests show certain The vertical transfer curves shown in the figure were obtained only when a striking anomalies. Two pipes, one definitely rougher than the other, were reversed length of pipe at least equal to 130 diameters is interposed between the entrance in their resistances when tested over a certain range of Reynolds’ number near the and the testing length. Under other conditions the transfer curve is less definite transfer region. This behavior was characteristic, in this region, of the majority of and becomes rounded in form and very similar to the dotted curve shown by the the pipes tested, and persisted long after the flow had ceased to be streamline. The 156 Chapter 3: The Design Revolution Document 3-7 157

conclusion may be drawn that no reliance can be placed on the relative magnitudes of the observed resistances of two models of a complicated form unless the comparison is made at Reynolds’ numbers many times as great as those of the transfer region, or, alternatively, unless the state of the fluid is completely known. In connection with pipe flow the writers have found logarithmic plotting along both axes to be more convenient than the usual method of plotting F/ρv2 on log. R. Equations (1) and (2) or the lecture would then become straight lines sloping downwards. Plotted as they are, there is a false impression given, owing to the upward concave form of the curves, that they are tending to some constant value of F/ρv2 at high values of R. There is a further advantage of logarithmic plotting in that inaccuracies of test data of equal percentage are consistently shown at all parts of the diagram by equal displacements, so that undue importance is not attached to isolated tests which happen to fall on parts of the diagram where the scale may be relatively close. Mr. SIMMONS (communicated): Professor Jones has tonight expounded a theory based on the assumption that the skin friction resistance of bodies of fine form follows one law at low values of Reynolds’ number and another law at high values. The underlying basis of the theory recalls the methods employed by naval architects for predicting the resistance of ships, founded on the skin friction data of flat plates. He would like to call attention to some results published some years ago by Mr. Baker which he thought pertinent to the present discussion, and lent support to the new theory. These were shown graphically in Fig. 6. It would be observed that below a Reynolds’ number of about 4 X 106 large discrepancies existed between REPLY TO DISCUSSION individual observations made at the same value of VL/v; although at higher num- The discussion has been long and interesting, and I wish first to thank all those bers the results were more generally consistent. Commenting on the measurements who have taken part in it. made in air and water on similar models, he thought it was significant that the drag In reply to Major F. M. Green, the turbulence in the wind tunnel which causes coefficient measured in the turbulent flow of the wind tunnel was higher than the early break up of the boundary layer is of a peculiar kind with a very small scale. I figure obtained in the water experiments. If the initial turbulence in the free stream imagine that the scale of the turbulence in the open air is too large to have any influ- was a factor which influenced the flow in the boundary and thereby the drag of the ence on the boundary layer and I should not therefore expect differences on differ- model, then its effect appeared to be most marked below ent days due to this cause. This opinion is not based on definite evidence. The figure given for skin friction drag coefficient is half the minimum profile drag coefficient, 6 VL/v = 4 X 10 . because the area involved in the former is that of both upper and lower surfaces of the planes, whereas the area involved in the latter is one surface only. He was, however, a little doubtful whether a stable regime of flow independent With reference to Mr. R. McKinnon Wood’s remarks, I agree that it is probably of initial disturbances was completely established at this value of Reynolds’ number, that wind tunnels which have been deliberately given a small scale turbulence will since the Gottingen results exhibited by the lecturer, and shown by the dotted curve be used in the future, but this matter has not yet been clearly thought out so far as were in excess of the figures derived from tests on the largest planks. Finally, he I am aware. referred to the similarity existing between the results for bodies of fine form in the Mr. Fage’s statement that the Reynolds’ number at which the boundary layer transitional region and those obtained with pipes; and he thought it must be more breaks up on a circular cylinder is of the same order as that for a flat plate, is most than coincidence that the turbulence curve for flat plates could also be taken to interesting. The effect of curvature on the point of break up of the layer is a matter represent the resistance of pipes at high velocities. which requires much further research. I shall be interested to see his results for thin Joukowski sections when they become available. 158 Chapter 3: The Design Revolution Document 3-7 159

Mr. Walker’s estimate that the speed of his aeroplanes lies between 60 and 67 I agree with Mr. Douglas as to the importance of a clean external surface, but percent of the streamline speed agrees fairly well with my rougher estimates. Since it is just possible that small projections may have less influence on the full-scale, power varies approximately as V3, the realized speed of 70 percent of the streamline where the boundary layer is probably turbulent in any case, than on the wind tunnel speed corresponds to an expenditure of about three times the streamline power for model, where the projections may convert a laminar into a turbulent layer. a given speed. I am obliged to Mr. Walker for emphasizing the fact that the ratio of Mr. Irving’s suggestion that the boundary layer may continue laminar to higher actual to streamline power must not be regarded as an overall figure of excellence Reynolds’ numbers in respect of some shapes is very significant. We had already for the aeroplane. I had a warning to this effect in an early draft of the paper, but given some thought to this matter at Cambridge, but I left it out of this paper as inadvertently omitted it in the final text. What the factor does show is how much being too speculative. This, I hope, will be the subject of future research. power is being expended through defective streamlining. I agree with his remarks His remarks about his experiments on skin friction in 1919, in which the skin about Mr. Lanchester and the induced drag theory; all the same, it was Prandtl who friction coefficient fell on the curve for laminar boundary layer, are most interesting threw it into its present practical form. and I should like the reference. Dr. Stanton has also done experiments on the skin Mr. W. L. Cowley finds it hard to reconcile my lecture with the available results friction on the thin rings similar to napkin rings, and has found the same agree- which have been given by wind tunnels in the past. I do not understand his dif- ment. There is no doubt that when the layer is laminar the agreement of the skin ficulty. The majority of results which have been used extensively have not been for friction with Blasius’ solution is thoroughly established. perfect streamline bodies such as I am discussing. Also it must be remembered that Mr. Capon has clearly emphasized the main reason for the lecture, which was the majority of full-scale results refer only to total brake horse-power, and that scale to show that an estimated sum of all conceivable drag reductions is so large as to be effects on the various parts may be confused with variation in airscrew efficiency worth going for, even if the separate contributions may in themselves appear small. or interference between airscrew and body. Only in a few instances, where research Mr. Scott Hall has doubts about my screw efficiency theory. I am not surprised methods have been applied on the full-scale, do we know the drag independent of at this, as it is at present the weakest part of the paper and is as yet unsupported by the airscrew, and even then the whole drag and not that of the parts separately. crucial experiment. The main conclusions of the paper, however, stand without it. There is one way in which apparent agreement can be obtained in spite of the The assumption certainly is, as he says, that the interference of screw on body bal- existence of large scale effects; thus, the airship R.101 when tested on the model at a ances the effect of body on screw, except for the increased power loss due to skin 6 Reynolds’ number of 2 X 10 gave kF = 0.0010. This figure was almost exactly that friction in the slipstream. Stated in this way, the assumption is certainly startling, which we should now predict for the full-scale with a wholly turbulent boundary but if my argument is carefully followed I do not see how the conclusion can be layer at a Reynolds’ number of 3 X 108. The apparent agreement here is entirely avoided. It is to be emphasized, however, that the argument applies only to ideally fortuitous, for at intermediate Reynolds’ numbers the coefficient is much higher, as streamlined machines as defined in the paper. would be expected from the theory which I have advanced. In reply to Mr. Branson, the suggestions in this paper cannot be considered as In Mr. Cowley’s remarks upon pipes he appears to have forgotten that in a long on the same plane as the Carnot theory of heat engines, but the practical outcome pipe the boundary layer extends to the middle of the pipe. There is a relationship is similar—the provision of an ideal towards which to work. Whereas, however, the between the flow in pipes and the flow for a flat plate, but it is not so simple as Mr. Carnot cycle is a precise theorem, my paper is more in the nature of an exercise in Cowley appears to assume; it has been worked out by Professors Prandtl and Von approximations. Kármán (See my reply to Messrs. S. J. Davis and C. M. White for references). I agree with Mr. Dowty that the landing gear is probably the greatest difficulty I cannot understand Mr. Cowley’s statement that the pressure distributions in the way of perfect streamlining, and I am interested to hear of his efforts to tackle when resolved are equal to the whole drag; so far as I am aware this is incorrect for the problem. streamline shapes. In answer to Captain Geoffrey Hill, the air-cooled engine is at present a distinct So far from the conclusions of my paper leading to a loss of faith in wind stumbling block to progress in streamlining, but I am not sure that it will always be tunnel work, they will, I hope, increase its usefulness through removing some of the so. Recent experiments in cylinder cowling are most encouraging and it should be anomalies which have been well known for some years. remembered that, since the temperature of the cylinder is high, scientifically devised In answer to Mr. Pye, I have not included the Schneider Cup racers because, air-cooling systems, such as we have not yet got, may give rise to a relatively low being seaplanes with floats, they are not comparable with the other aeroplanes in resistance. Fig. 3. It would be interesting to see them worked out accurately by someone in the In reply to Messrs. S. J. Davies and C. M. Whyte, the relation between resistances Air Ministry who has the facts at his fingers’ ends. to flow in pipes and the Reynolds’ number formed from the pipe diameter was 160 Chapter 3: The Design Revolution Document 3-7 161

thoroughly worked out by Stanton and Pannel for both water and air, and a curve assumption regarding the point of breakdown from laminar to turbulent flow is that showing the relation of skin friction per unit area/ρV2 to the Reynolds’ number, for the point is uninfluenced by the portions of the plate behind it. This assumption a very wide range including the critical region, was published in 1914. must surely be correct for the flat plate; whether or not it is correct for the curved

The comparison between flow in pipes and the boundary layer on a flat plate surfaces is shown by comparing the curves for kF against R for the various streamline has been thoroughly examined by Von Kármán, of Aachen (Abhandlunger aus dem bodies with the theoretical transition curves for the flat plate. Aerodynamischen Institut an der Technischen Hochschule, Aachen 1921) and by To avoid unduly lengthening the paper, I did not give my methods of calculat- Prandtl (Ergebnisse der Aerodynamischen Versuchsanstalt zu Gottingen, 1927, Vol ing the transition curves. The difficulty lies in deciding to what extent the laminar III). Both these papers have been translated into English, the former by the A.R.C. layer on the front part of the plate will influence the turbulent layer behind it. If it is assumed not to influence the turbulent layer at all, an upper limit is given for the as Report T. 2219 and the latter by the Air Ministry. In these papers a law, kF varies as (lυ/v)-0.20 for flat plates with turbulent boundary layer is, with certain plausible coefficient of the average drag of the whole plate. If, on the other hand, it is assumed -0.25 that the turbulent part in the rear behaves as though the whole layer were turbulent, assumptions, deduced from the law kF varies as (lυ/v) which is found to hold experimentally for pipes above the critical Reynolds’ number. Fig. 7 is a composite a lower limit is given to the average coefficient. In forming the transition curves in diagram taken from two reports in the Gottingen publication mentioned above and Figs. 1 and 2 I worked out curves on both the above assumptions and took the mean -0.2 between them. Prandtl, on the other hand, worked on the second assumption only roughly re-plotted to conform with Figs. 1 and 2. The curve kF = 0.037(iυ/v) adopted by Prandtl, though it fits Gebers’ points at the higher Reynolds’ numbers, in producing his transition reproduced curve in Fig. 7. The difference between my falls lower than the results from Froude and the N.P.L. given in Fig. 7. Hence for method and Prandtl’s is not great, and I do not know which is nearer the truth. practical purposes I prefer his original formula The question of the effect of curvature on the stability of the boundary layer is -0.15 still obscure, and will, I hope, shortly be the subject of experimental investigation. kF = 0.019(lυ/v) , which gives a conservative estimate of skin friction drag for flat plates over the whole known range. I am aware of the advantage of complete logarithmic plotting, but I thought I do not understand the writer’s difficulties with the transition curves. The only that on the balance the system which I adopted would be more easily explained to 162 Chapter 3: The Design Revolution

members of the audience who might not be very familiar with logarithmic plotting. The use of logarithms in connection with Reynold’s numbers is easily explained as a convenient device for compressing a large range of numbers into a small space. I am much obliged to Mr. Simmons for drawing attention to the available data contained in the figure accompanying his remarks. My attention had been drawn to the existence of this data by Sir Richard Glazebrook a day or two before the lecture, but I had not had an opportunity to consider it. Since the lecture I have secured Mr. Baker’s original paper and have re-plotted the information on the same plan as Figs. 1 and 2 though on a larger scale, and the resulting diagram is reproduced herewith (Fig. 8). It is evident, both from Mr. Simmons’ figure and from mine, that the boundary layer in the N.P.L. tank experiments was breaking up at about R = 6 X 105, and it is clear from my figure that at high Reynolds’ numbers the curves of Froude and the N.P.L., which continue up to 4 X 107, lie within some 5 percent of -0.15 the equation kF = 0.019(lυ/v) . These latter data bring our knowledge of the skin friction coefficient of a flat plate up to a Reynolds’ number of 4 X 107, that is to say, well into the region of Reynolds’ numbers represented by full-scale aeroplane wings and bodies. They show that the formula which I used in my lecture for the skin friction is on the safe side in this region. Document 3-8 163

Document 3-8

Excerpts from Fred E. Weick and James R. Hansen, From the Ground Up: The Autobiography of an Aeronautical Engineer (Washington and London: Smithsonian Institution Press, 1988), pp. 49-61, 66-68, 72.

The organizing thinker and team leader of the NACA’s original cowling program at Langley was Fred E. Weick, one of the most remarkable aeronautical engineers in the history of American aeronautics. Born near Chicago in 1899, Weick (pro- nounced Wyke) developed an avid interest in aviation by the age of 12, attending air meets at nearby Cicero Field and entering model airplane competitions. Upon graduation from the University of Illinois in 1922, he began his professional career as a draftsman with the original U.S. Air Mail Service. After a short stay with the Yackey Aircraft Company (during which time he worked in a converted beer hall in Maywood, Illinois, transforming war-surplus Breguet biplanes into “Yackey Trans- ports”), he started a job with the U.S. Navy Bureau of Aeronautics in Washington, D.C., where, within a matter of months, the NACA’s director of research, George W. Lewis (1882-1948), personally recruited him for important work to be done at Langley, some 120 miles to the southeast. (The NACA’s Washington office was located in an adjacent wing of the Navy Building, thus facilitating close relations between the NACA and the navy.) Weick arrived at Langley in November 1925 just in time to take over the design and construction of the new Propeller Research Tunnel (PRT)—the job Lewis had specifically asked him to do. The following is a series of excerpts from Fred Weick’s 1988 autobiography in which he recalled the construction of the PRT and the origins of the cowling research program more than 60 years earlier. Readers will find that Weick’s recall of these events from long ago was amazing. Perhaps even more amazing was that Weick was such a marvelously clear thinker when it came to technology and how his approach to everything related to engineering was strictly rational. One of Weick’s friends and colleagues at NACA Langley in the 1930s (Weick worked at Langley from late 1925 to 1929, then moved to a job with Hamilton Aero Manufacturing Company in Milwaukee, Wisconsin, maker of adjustable aluminum-alloy propellers and steel hubs for both military and commercial aircraft, and then returned to work at Langley from 1930 to 1936) was the distinguished aerodynamicist Robert T. Jones, who became one of the fathers of the swept wing. In the foreword to Weick’s autobiography, Jones paid tribute to this former NACA associate specifically in terms of his role in “reinventing the airplane.” “Working with Fred,” Jones wrote, “I had the feeling that the airplane was being reinvented, as it indeed was. Never mind that the general form and arrangement of the airplane had been well 164 Chapter 3: The Design Revolution Document 3-8 165 established for many years, Fred felt strongly that every function of the airplane construction of the variable-density tunnel. I was placed under Miller until the needed to be studied from a logical point of view, without prejudice, so that design- tunnel was ready for operation. ers and operators could make whatever changes were necessary to improve those By the time I started at Langley, the outer shell of the new tunnel had been functions” (p. viii). completed but work on the entrance and exit cones and guide vanes was still going Following his pioneering work on the NACA cowling, Weick went on to make on. The tunnel, which had been laid out by Dr. Max Munk in Washington, was of many other significant contributions to the advancement of aeronautical technol- the open-throat type then most suitable for testing propellers. My first job was to ogy, including development of the steerable tricycle landing gear, the conventional design and get constructed a balance arrangement that measured the aerodynamic gear used today—even for the Space Shuttle. His most widely recognized achieve- forces on the model and the model’s reaction to them. This balance had to support ment, the Ercoupe, has been the favorite airplane of thousands of private flyers since an airplane fuselage, complete with engine and propeller, 25 feet above the floor in its first production model came out in 1940. And his revolutionary AG-1 and Piper the center of the tunnel’s 20-foot-in-diameter airstream. All of the pertinent forces, Pawnee set life-saving standards of lasting benefit to both the agricultural airplane such as drag, thrust, and moments, were to be measured down below by four small and general aviation industries. His autobiography from which these excerpts are and simple beam scales. taken tells his entire life story in fascinating detail, from his days with the barn- Since 1921 Dr. Munk had been holed up in a little office at NACA headquarters stormers, through his navy and NACA years, to his many years in manufacturing in Washington, where he had been turning out excellent theoretical work. Munk for the Engineering Research Corporation (ERCO) and Piper. had studied under Ludwig Prandtl at the University of Gottingen in Germany and had been brought to this country by the NACA in 1921. His entry into this coun- Document 3-8, Excerpts from Fred E. Weick and James R. Hansen, From the try required two presidential orders: one to get a former enemy into the country, Ground Up: The Autobiography of an Aeronautical Engineer (Washington and and another to get him a job in the government. And I guessed this helped him to London: Smithsonian Institution Press, 1988), pp. 49-61, 66-68, 72. appreciate his importance. Without question, Munk was a genius, and, without question, he was a dif- National Advisory Committee for Aeronautics, ficult person to work with. In early 1926 he decided on his own that, since Langley Langley Field laboratory was where all the real action was taking place, that was where he should be. NACA headquarters must have agreed, because it made him the lab’s chief of The distance by air from Washington, D.C., to Hampton, Virginia, the town aerodynamics; this put him in charge of the flight research division and the two nearest Langley Field, is only about 120 miles, but by road through Richmond it is wind tunnel sections. My boss, Elton Miller, now reported to Munk, and all of my about 175 miles. In the 1920s these roads were surfaced with gravel and often badly work ultimately had to be approved by him. I had known Munk in Washington and rutted; smooth ribbons of concrete were not to be found in rural Virginia. In our had great respect for his abilities. On the other hand, I did not want my balance Model T roadster, packed to overflowing, it took us all day to make the trip. design turned down at the last minute; so I had taken the pain to take each detail The engineer in charge of the Langley Memorial Aeronautical Laboratory at of design, mostly on cross-section paper, up to Munk to get his approval, and I got that time, a Californian by the name of Leigh M. Griffith, appeared unhappy with his initials on every single one of them. This, I thought, would certainly assure his the idea that I had been placed under him from above; in fact, Griffith must have final approval. been generally unhappy with his situation at Langley, for he left within the month. The movable parts of the balance supporting the airplane were supported by He was replaced by Henry J. E. Reid, an electrical engineer who had been in charge a structural steel framework about 12-feet high, 12-feet wide, and 16-feet long. of the laboratory’s instrumentation. Reid remained the engineer-in-charge until he In place of adjustable cables, steel angles ¼ by 2 ½ by 2 ½ inches provided the retired from the National Aeronautics and Space Administration (NASA) in 1960. diagonal bracing. The lab had a flight research division headed by test pilot Thomas Carroll, a power A couple of days before we expected to try out the balance using a little Sperry plant division headed by Carlton Kemper, and two wind tunnel sections, one, the Messenger airplane with its 60-horsepower engine running, Munk made an unan- 5-foot or atmospheric wind tunnel (AWT) section headed by Elliott G. Reid, and, nounced visit to the PRT building. Just as he walked into the bare-walled 50-foot the other, the variable-density tunnel (VDT) section headed by George J. Higgins. cubicle that housed the test section, a loud horn squawked, calling someone to the There was also an instrument shop, model shop, technical service department, and telephone. This sent Dr. Munk into a tantrum, and I immediately had one of my a clerical and property office headed by Edward R. (“Ray”) Sharp. The new 20-foot mechanics disconnect the horn. Before he had entirely calmed down, he walked propeller research tunnel (PRT) was being constructed under the supervision of over toward the balance structure and put his hands on the long diagonal braces. Elton W. Miller, a mechanical engineer who had previously been in charge of the 166 Chapter 3: The Design Revolution Document 3-8 167

These were fairly flexible, and he found he could move them back and forth a bit. In 1926 Dr. Munk gave a number of lectures on theoretical aerodynamics to a Visualizing the entire structure vibrating to the point of failure and the whole air- select group of young Langley engineers. I was very happy to learn these things from plane and balance crashing to the ground, the perturbed Munk ordered me to tear him. Ever since graduation from the University of Illinois, I had thought about down the balance entirely and to design a new foundation and framework for it. He taking some graduate courses in aeronautical engineering. While working for Tony then turned and went back to his office a couple of blocks away. Yackey, I had read in a magazine article about the graduate courses in aeronautics Naturally I, too, was perturbed. Munk, after all, had approved every detail of offered at Massachusetts Institute of Technology. I had written MIT for informa- my balance design. Not knowing what to do, I waited for some time to give him tion and had received a letter back from Professor Edward P. Warner, who had been an opportunity to cool down. Then I went to his office and, as calmly as I could Langley’s first chief physicist in 1919 and would later become assistant secretary of manage, mentioned that I thought the natural frequencies of the long diagonal the navy for aeronautics, editor of the magazine Aviation, and finally president of members would be so low that vibrations would not be incited by the more rapid the International Civil Aviation Organization (ICAO), which continues to coordi- impulses from the engine and propeller. But mainly I suggested that, inasmuch as nate the rules and regulations for aeronautical activities throughout the nations of all the parts were made and ready to be put up, why not wait a couple of days before the world. I had hoped still to find a way to work in some graduate courses even tearing it down and make a careful trial using the Sperry Messenger, starting at after reporting to work at the Bureau of Aeronautics. But Dr. Lewis had talked me low speed, gradually increasing it, before dismantling the apparatus. Munk finally out of the idea on the basis that formal aeronautical engineering education was agreed, but demanded to be present when the test was made. inferior to what I could learn if I went to work for the NACA at Langley. I guess he I did not like that idea of his presence one iota. To start the engine, the Messen- was probably right in regard to the aeronautical courses per se, but on occasion, in ger’s propeller had to be cranked by hand from a balloon ladder that was put up in later years, I sorely missed the extra mathematics and physics that would have been front of the propeller 25-feet above the floor. (A balloon ladder was like a fireman’s obtained in school. ladder but its base was attached to a pair of weighted wheels, which permitted it to As mentioned earlier, the power plant for the new PRT consisted of two 1,000- be “leaned” out into space. At its base there was a “protractor” that told you how far horsepower, six-cylinder in-line diesel engines taken from a T-2 submarine. These it could be angled without tipping over.) This sweaty business often took some time. engines were located end-to-end with crankshafts connected to a large sheave or It was not the kind of operation I wanted the excitable Munk to watch. Moreover, pulley between them. This sheave carried forty-four Tex-rope V-belts to a similar since no one else in the PRT section had ever started an airplane engine by turning sheave on the shaft of the propeller fan that drove the air through the tunnel . The the propeller, I was the one who was going to have to do it. shaft of the propeller fan was 25 feet above the ground, and the two sheaves were 55 I brought my problem to Elton Miller, my boss, and to Henry Reid, the engi- feet apart center to center. Because we were concerned that some destructive vibra- neer-in-charge. Together, we decided that the only thing to do was to make an end- tions might occur in the crankshaft-sheave assembly, we decided that a theoretical run around Munk and check out the tunnel balance system in his absence. This was analysis of the torsional oscillations should be made, with Dr. Munk outlining the easily done, as Munk worked on theoretical problems in his room at a Hampton problem and a new man, Dr. Paul Hemke, to work out the solution. As a junior boarding house every afternoon. We set up the test run and after a bit got the engine engineer, my assignment was to give the measurements and sizes that I would get started without any difficulty. We then experimented with it until we could start it from the drawings of the engines and sheaves. easily and felt ready for the final trial. I had no difficulty giving them the measurements, but Dr. Hemke was never The problem of convincing Munk remained. We could not simply tell him able to get the gist of the torsional pendulum problem as described by Munk. This about the successful test, so we agreed to arrange another “first test” for Munk to went on for some time with no results being obtained. Finally, I looked into my witness. Engineer-in-charge Reid escorted Munk to the tunnel the next morning. mechanical engineers’ handbook and into a couple of textbooks and found that I casually said, “Good morning,” clambered up the ladder, and pulled through the considerable work had been done on the problem and that the solution was not Messenger’s prop. Luckily, the engine started on the first try. We then moved the too difficult. I made the computation myself, coming out with a natural frequency ladder away, ran the engine through its entire range with no vibration difficulty, and of 312 RPMs. Later on, after the tunnel was in operation, some men came down then shut it down. Now, I wondered, what sort of explosion will we have? I needn’t from the navy shipyard in Brooklyn with equipment to measure the torsional oscil- have worried. Munk walked toward me with his hand outstretched and congratu- lations; they found exactly the same natural frequency as I had computed. Hitting lated me on the success of the operation. Everything had turned out all right. The it exactly, of course, was a matter of luck, but it helped give me a good reputation, balance system of the PRT operated satisfactorily with engines of up to 400 horse- whether I deserved it or not. The success put me in good with Munk, but unfortu- power into the late 1930s, when it was replaced by a new and better one. nately Dr. Hemke was never able to work satisfactorily with him. A short time later 168 Chapter 3: The Design Revolution Document 3-8 169 he left the NACA. Hemke later joined the faculty of the U.S. Naval Academy, after that was going on in them. The propeller research tunnel was about finished, but holding a prestigious Guggenheim Fellowship for research under B. Melville Jones Ted Myers, who was in charge of the tunnel’s power plant, had not been able to at Cambridge. get the diesel engines to run. However, we had the regular starting arrangement by Another problem I helped to solve was the design of the 28-foot propeller fan which we turned the engines over by a blast of compressed air until they would start that was to circulate the air in the propeller research tunnel. This fan needed to have running as diesels. At the demonstration that morning, we ran the tunnel on the eight blades of normal width. The exact energy ratio of the tunnel was not known compressed air for about one minute; the little Sperry Messenger was up in the test in advance, so I desired to have blades that could be adjusted so that the pitch could section with its engine running also. In the afternoon, the conference was held at the be set exactly right after trial runs. Aluminum-alloy blades therefore seemed the military officers’ club a few blocks away, and suggestions for possible new research best choice, but the blades we wanted were too large to be forged in the manner were invited. One of the suggestions that was made concerned the cowling of radial of the aluminum-alloy propeller blades then being manufactured. Fortunately, the air-cooled engines. propeller was to turn at only 375 revolutions per minute, which meant that the When I had started work at the Bureau of Aeronautics, almost all of the army stresses would be very low in comparison even with airplane propellers having large and navy airplanes had had water-cooled engines. The navy, however, was interested diameters. This gave me the idea that a cast aluminum alloy might be used success- in developing radial air-cooled engines. This work had been carried on under the fully, which it was. direction of Comdr. Eugene E. Wilson of the bureau’s power plant division and had I arranged with the Aluminum Company of America to cast the blades in their been conducted mostly with the Wright engines designed by Charles Lawrance. The plant at Cleveland, Ohio. Before the large blades were cast, however, the company radial engines with their short crankshafts and crankcases and no radiators or water- made two blades for a small ten-foot model that I then took to McCook Field in cooling systems were lighter than the water-cooled engines. But the finned cylinders Dayton, where they were tested by Army Air Service engineers on their propeller were cooled simply by projecting them into the airstream, and this caused a high whirl rig. This test showed the blades to be sufficiently strong. drag. An attempt had been made to reduce the drag by putting propeller spinners During the period that the blades were being manufactured, I made a number over the hubs and cowling the crankcase and lower portions of the cylinders, but the of trips to Cleveland. On one occasion, when I had an afternoon with nothing outer ends of the cylinders still extended into the airstream. special to do, I visited the Martin aircraft plant in the city’s southern suburbs. I During the morning session of the NACA conference, everyone had witnessed went into the door to the main office and told a young lady at the desk that I was the operation of the Sperry Messenger airplane, with its radial air-cooled Lawrance an engineer from the NACA at Langley Field and that I’d like to visit the plant. She engine running, in the propeller research tunnel. At the afternoon meeting, several ushered me into the office of Glenn L. Martin himself, and he spent a couple of people mentioned that tests should be made in the PRT to see how much the cowl- hours showing me around. How different from the stilted, bureaucratized condi- ing could be extended outward without interfering too much with the cooling of tions existing today in an aircraft factory! Of course, the Martin plant was small the engine. Both the drag and propeller efficiency should be determined, we all then, with only a few hundred employees. Most of his production went to the Navy agreed, as well as the cooling . During the ensuing months, I laid out a program for Department; in fact, while at the Bureau of Aeronautics, I had designed a couple these cowling tests. of the propellers used on his airplanes. Because many of his models were seaplanes While studying propellers at the Bureau of Aeronautics, I learned from the and nearby Lake Erie was frozen solid in the winter months, Martin was then look- propeller work carried out by William F. Durand and Everett P. Lesley at Stanford ing around to find a place farther south where he could manufacture and fly them University the advantages of using a systematic series of independent variables in away directly from the factory all year round. On this account he asked about the experimental research . I recognized that the range of variables should extend, if conditions around Hampton and Newport News, a neighborhood that he thought possible, on both sides well beyond the area of greatest interest. One extreme of this might be quite suitable. I told what I could about the area, and later he made some series was obviously making use of the bare engine with no cowling at all. The other overtures in this direction. But, as I remember it, the local people at Newport News extreme was to enclose the engine completely. This option had not been anticipated were not interested. Martin finally moved his factory to Middle River, Maryland, but looked enticing. An engine nacelle would then start with the best airship shape near Baltimore, where local authorities gave him a very good deal. available, and the air could be brought in smoothly at the center of the nose. But The NACA held its first annual manufacturers’ conference at Langley in May how could one get the air out again in a smooth and efficient manner? Elliott G. 1926. The meeting was attended by representatives from the military air services, Reid, who was in charge of the atmospheric wind tunnel at the time, had been Department of Commerce, and aeronautical manufacturing industry. The morning making tests on Handley Page wing slots, and he helped me to design an annular was spent touring the various laboratories and learning about the research work exit slot. Together, these forms eventually became the NACA’s low-drag cowling. 170 Chapter 3: The Design Revolution Document 3-8 171

After I had completed the outline of a tentative cowling test program, the propulsive efficiency determined with a metal propeller. The propulsive efficiency NACA sent it to the military air services and to various manufacturers that had was found to be practically the same with all forms of cowling. The drag of the cabin shown interest at the May 1926 conference, and it was approved by all of them. For- fuselage with uncowled engine was found to be more than three times as great as the tunately, getting their okay took some time, because the propeller research tunnel drag of the fuselage with the engine removed and nose rounded. The conventional was at this point in no sense ready to operate. forms of cowling in which at least the tops of the cylinder heads and valve gear are After establishing that the tunnel was operating satisfactorily, we carried out exposed, reduced the drag somewhat, but the cowling entirely covering the engine several series of propeller tests and cowling tests at the same time. Among other reduced it 2.6 times as much as the best conventional one. The decrease in drag due things, this enabled us to obtain the effect of propeller-body interference on each to the use of spinners proved to be almost negligible. cowling design. The various propeller tests were mostly covered in the following I concluded this summary by arguing that use of the form completely covering NACA reports: TR 306, “Full-Scale Wind-Tunnel Tests of a Series of Metal Propel- the engine was “entirely practical” under service conditions, but also by warning lers on a VE-7 Airplane” (July 13,1928); TR 338, “The Effect of Reduction Gearing that “it must be carefully designed to cool properly.” on Propeller-Body Interference as Shown by Full-Scale Wind-Tunnel Tests” (March Having completed the initial round of wind-tunnel tests, we then borrowed a 20, 1929); TR 339, “Full-Scale Wind-Tunnel Tests with a Series of Propellers of Curtiss Hawk AT-5A airplane from the Army Air Service at Langley Field already Different Diameters on a Single Fuselage” (March 12,1929); TR 340, “Full-Scale fitted with the Wright Whirlwind J-5 engine, and applied the new cowling for Wind-Tunnel Tests on Several Metal Propellers Having Different Blade Forms” flight research. These tests showed that the airplane’s speed increased from 118 to (March 18,1929); and TR 350, “Working Charts for the Selection of Aluminum 137 miles per hour with the new cowling, an increase of 19 MPH. The results of Alloy Propellers of a Standard Form to Operate with Various Aircraft Engines and the instrumented flight tests had a little scatter, and we could have been justified Bodies” (March 25, 1929).1 in claiming that the increase in speed was 20 MPH instead of 19, but I wanted to The goal that we had set for ourselves in the cowling program was a cowled be conservative. I didn’t want people to expect too much from this cowling, so we engine that would be cooled as well as one with no cowling whatsoever. This program called it 19. proceeded easily enough until the complete cowling, covering the entire engine, was The second part of the cowling program covered tests with several forms of first tried. At this point, some of the cylinder temperatures proved to be much too cowling, including individual fairings behind and individual hoods over the cylin- high. After several modifications to the cooling air inlet and exit forms, and the use ders, and a smaller version of the new complete cowling, all mounted in a smaller, of internal guide vanes or baffles, we finally obtained satisfactory cooling with a open-cockpit fuselage. We also performed drag tests with a conventional engine complete cowling. Don Wood was in charge of the actual operation of the testing, nacelle and with a nacelle having the new complete design. Though the individual and the first of these modifications was made while I was away on a vacation. When fairings and hoods proved ineffective in reducing drag, we found that the reduction I got back, it was obvious to me that the boys were on to something, and from that with the complete cowling over that with the conventional cowling was in fact over time on we all worked very hard on the program. twice as great with smaller bodies as with the larger cabin fuselage. Data from the The results of this first portion of cowling tests were so remarkable that we AT-5A flight tests confirmed this conclusion. decided that the NACA should make them known to industry at once. In Novem- The first public acclaim of the cowling came in February 1929 when Frank ber 1928 I wrote up Technical Note 301, “Drag and Cooling with Various Forms Hawks established a new Los Angeles-to-New York nonstop record (18 hours, 13 of Cowling for a ‘Whirlwind’ Engine in a Cabin Fuselage,” which the NACA pub- minutes) flying a Lockheed Air Express equipped with an NACA low-drag cowling lished immediately . The summary of the report was as follows: that increased the aircraft’s maximum speed from 157 to 177 MPH. The day after The National Advisory Committee for Aeronautics has undertaken an inves- the feat, the NACA received the following telegram: tigation in the 20-foot Propeller Research Tunnel at Langley Field on the cowling Cooling carefully checked and OK. Record impossible without new cowling. of radial air-cooled engines. A portion of the investigation has been completed in All credit due NACA for painstaking and accurate research. [Signed] Gerry Vultee, which several forms and degrees of cowling were tested on a Wright Whirlwind J-5 Chief Engineer, Lockheed Aircraft Co. engine mounted in the nose of a cabin fuselage. The cowlings varied from the one Some time later, the NACA gave me a photographic copy of this telegram, extreme of an entirely exposed engine to the other in which the engine was entirely along with a picture of the cowled airplane. enclosed. Cooling tests were made and each cowling modified if necessary until the A few weeks before Hawks’s record-breaking flight, I had attended the New engine cooled approximately as satisfactorily as when it was entirely exposed. Drag York Air Show in Madison Square Garden. At the show Chance Vought told me tests were then made with each form of cowling and the effect of the cowling on the that Germany’s Claude Dornier would like to talk to me about the possibility of 172 Chapter 3: The Design Revolution Document 3-8 173 putting the NACA cowling on the twelve uncowled radial air-cooled engines of efficiency of the Fokker airplane made it clear that the optimum location of the his giant DO-X flying boat—which, at the time, was the largest airplane in the nacelle was directly in line with the wing, with the propeller well ahead of the world. These twelve engines (British-made Jupiters rated at 550 horsepower each) wing’s leading edge. This position had the least overall projected area, and I suppose were mounted back-to-back in six nacelles, each with one tractor propeller and one the result might have been expected. With the complete cowling, the radial engine pusher propeller. Cowling the pushers would, of course, constitute an entirely new in this position spoiled the maximum-lift coefficient of the wing. With the cowl- problem. ing, and the smooth airflow that resulted from it, the maximum-lift coefficient was Before finishing this story, it should be mentioned that a great effort was then actually increased. After this important information was transmitted confidentially being made in a number of countries to develop aircraft suitable for airline use to the army, navy, and industry, most all of the transport and bombing airplanes across the oceans, particularly the North Atlantic. The aircraft were of two main employed radial wing-mounted engines with the NACA cowled nacelles located types: rigid airships of the Zeppelin type and large flying boats like Dornier’s. The approximately in the optimum position. wing of the DO-X projected from the top of an ample hull with a span of 157 feet and a chord of 30 feet. Lateral stability on the water was obtained by the use of sponsons, or short and stubby winglike structures that projected from the bottom of the hull on each side. Constructed in the late 1920s at Altenrein, Switzerland, on Lake Constance and near Friederichshafen, Germany, the DO-X could accommodate sixty-six passengers comfortably over a range of 700 to 900 miles, but could not lift any kind of payload over transatlantic distances, the minimum such distance being roughly 2,000 miles. On one flight from adjoining Lake Constance, though, one hundred seventy people were crowded into the airplane (I’ve heard that nine were stowaways), making quite a record at that time. By this time at Langley lab, we had mounted the low-drag cowling on all three engines of a Fokker trimotor airplane. The comparative speed trials proved extremely disappointing. Separate tests on the individual nacelles showed that cowling the Fokker’s nose engine gave approximately the improved performance we expected . Cowling the wing nacelles, however, gave no improvement in performance at all. This was strange, because the wind-tunnel tests had already demonstrated convinc- ingly that one could obtain much greater improvement with a cowled nacelle than with a cowled engine in front of a large fuselage. Some of us started to wonder how the position of the nacelle with respect to the wing might affect drag. In the case of the Fokker (as well as the Ford) trimotor, the wing engines were mounted slightly below the surface of the wing. The upper surface of the fully cowled nacelle then came very close to the under surface of the wing. As the air flowed back between the wing and nacelle, and the distance between them increased toward the rear of the nacelle, the expansion required was too great for the air to follow smoothly. We tried fairing-in this space, but achieved only a small improvement. As a result of these experiences I laid out a series of model tests in the propeller research tunnel in which an NACA-cowled nacelle with a power-driven propeller was placed in a number of different positions with respect to the wing. Where it appeared pertinent, extra fairing was put between them. These tests were run by Don Wood and his crew after I had left the NACA to work for Hamilton Standard. The resulting data on the effect of the nacelle on the lift, drag, and propulsive

Document 3-9(a-b) 175

Document 3-9(a-b)

(a) Western Union telegram, Jerry Vultee, Lockheed Aircraft Co., Burbank, CA, to NACA, “Attention Lieutenant Tom Carroll,” 5 Feb. 1929, copy in NACA Research Authorization file 215, NASA Langley Historical Archives.

(b) Fred E. Weick, “The New NACA Low-Drag Cowling,” Aviation 25 (17 November 1928): 1556-1557, 1586, 1588, 1590.

National attention became focused on the success of the NACA’s low-drag engine cowling in February 1929 when celebrated pilot Frank Hawks set a new nonstop speed record from Los Angeles to New York in a Lockheed Air Express equipped with a NACA cowl that increased its top speed from 157 to 177 miles per hour. Gerald “Jerry” Vultee, chief engineer with the Lockheed Aircraft Company, sent a telegram to NACA Langley’s chief research pilot, Thomas Carroll, crediting the NACA for the flight’s success; this telegram can be found as the first document below. Not only did the low-drag cowling provide unprecedented performance for new aircraft like the Air Express, it was also a factor in generating public notoriety for how much U.S. aviation was advancing generally. It also bolstered the reputation of the young NACA. Amid a burst of publicity—some of it exaggerated—about the benefits of the NACA cowling, the National Aeronautics Association announced in January 1930 that the NACA cowl had won the Collier Trophy for the greatest achievement in American aviation in 1929. The second document provides NACA Langley engineer Fred E. Weick’s article on the low-drag engine cowling published in Aviation magazine in November 1928. This was still early in the NACA’s cowling program. Work on advanced forms of cowlings continued throughout the 1930s. 176 Chapter 3: The Design Revolution Document 3-9(a-b) 177

Document 3-9(a), Western Union telegram, Jerry Vultee, Lockheed Aircraft Co., Document 3-9(b), Fred E. Weick, “The New NACA Low-Drag Cowling,” Burbank, CA, to NACA, “Attention Lieutenant Tom Carroll,” 5 Feb. 1929, copy Aviation 25 (17 November 1928): 1556-1557, 1586, 1588, 1590. in NACA Research Authorization file 215, NASA Langley Historical Archives. STATIC radial air-cooled engines are claimed by some engineers to possess WESTERN UNION several advantages over engines of other types, among which are low weight per horsepower, the small number of parts required, and the consequent reliability and FEB 5, 1929 low cost of manufacture. They are now widely used and are still gaining favor, both in commercial and military aeronautics. They have had, however, one outstanding To: NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS disadvantage--their extremely high air resistance, caused by the large frontal area and very poor aerodynamic form. The high drag has been a serious handicap in the Street and No.: LANGLEY MEMORIAL AERO LABORATORY LANGLEY field of high speed pursuit planes and single seater fighters, but the low weight of FIELD VIR the air-cooled radial has some advantages. Aircraft designers have attempted to reduce the drag due to the air-cooled radial Place: engine by enclosing varying amounts of it within the nose of the fuselage or nacelle. There has been a wide difference of opinion regarding the best forms and amounts ATTENTION LIEUTENANT TOM CARROLL LOCKHEED AIR of cowling, and its effect on the drag and the cooling of the engine. Some designers EXPRESS PILOT: FRANK HAWKS WHICH JUST ESTABLISHED NEW have cowled in almost the entire engine leaving only the cylinder heads and valve LOS ANGELES TO NEW YORK NONSTOP RECORD OF EIGHTEEN gear exposed, while others have been content to leave the engine entirely exposed. HOURS THIRTEEN MINUTES IS EQUIPPED WITH N.A.C.A. COWLING Practically no accurate or reliable information has been available on the subject, for WHICH INCREASED ITS SPEED TWENTY MILES PER HOUR FROM until very recently there has been no practical way of obtaining it. 157 TO 177 STOP COOLING CAREFULLY CHECKED AND OK STOP The National Advisory Committee for Aeronautics has made a practice for each RECORD IMPOSSIBLE WITHOUT NEW COWLING ALL CREDIT DUE of the last few years of inviting the aircraft manufacturers and their representatives to N.A.C.A. FOR PAINSTAKING AND ACCURATE RESEARCH AND GEN- spend a day at the Committee’s laboratory at Langley Field, in order that they may EROUS POLICY STOP KINDEST PERSONAL REGARDS STOP LETTER become more familiar with the work and facilities, and make suggestions regarding WITH DATA AND PHOTOS FOLLOWS. research which would be helpful to them. At the meeting held on May 24, 1927, the manufacturers were practically unanimous in urging that an investigation be JERRY VULTEE made in the new 20 ft. propeller research wind tunnel, then just being completed, LOCKHEED AIRCRAFT CO. on the cowling of radial air-cooled engines. The new tunnel, in which an air speed of 110 m.p.h. can be obtained, is ideal for the purpose. An actual full size airplane can be accommodated except for the wing tips, and the engines can be run with its propeller as in flight. The propeller thrust, airplane drag, and propulsive efficiency can be measured under flight conditions but with laboratory accuracy, so that small differences due to slight changes in cowling are brought out, and thermocouples can be used to obtain the temperatures at a large number of points on the cylinders so that the effect of the cowling on the cooling can be studied. In working out a program for the cowling tests, it was thought desirable to include not only all of the main conventional forms of cowling, but also to have them arranged in series with various degrees or amounts of cowling. At one extreme of the series, the engine was left entirely exposed except for the rear of the crankcase where it was fitted to a fuselage. For the other extreme, it seemed logical to enclose the entire engine. This involved problems in design, for it was of course desired to have the drag as low as possible, but still have the engine cool satisfactorily. It 178 Chapter 3: The Design Revolution Document 3-9(a-b) 179

was easy to design a form enclosing the engine and having a low drag, but there and nose rounded was more than tripled by the addition of the engine, and that of was no information available as to the best means of cooling the engine. It was the open cockpit fuselage was increased nearly five times. In fact, the drag of the decided to make the outside of the complete cowling of circular cross-section and open cockpit fuselage was about 10 percent. greater than that of the cabin fuselage smooth form, without individual cylinder fairing such as have been used on a few with the same engine cowling even though the open cockpit fuselage had only half experimental airplanes, particularly in England. This was done partly for the sake the cross-sectional area of the other. The larger body behind the engine evidently of simplicity and partly because in the case of most American radial engines, and has a beneficial effect on the drag, and the drag of a small fuselage without engine is especially the Wright Whirlwind J-5 which was used in this investigation, there is insignificant compared with that when the engine is added. The conditions with a so little room between adjacent cylinders that individual hoods are impractical. It wing engine nacelle would probably be at least as bad as with a small fuselage. These was necessary, of course, with the cowling entirely covering the engine, to separate facts showed plainly that there was an opportunity for great improvement, especially the air used for cooling the cylinders from the general flow about the body, guide it in the cases of small fuselages and engine nacelles. past the cylinders, and then return it to the outside air again. If the separation and The next outstanding development of the tests was that the conventional cowl- the return could be made smoothly, it seemed likely that a large decrease in drag ings, in which only the central portion of the engine is covered, had but slight effect could be obtained. in reducing the drag. Even in the extreme case, in which a large spinner was used It was decided that the cooling air for all of the cylinders could be taken in most and the entire engine was covered in except for the tops of the cylinder heads and satisfactorily at the center of the nose where the air pressure on the body, when in valve gear, the reduction in drag was comparatively small. Spinners as convention- motion, is greatest. This allowed a simple symmetrical design and a smooth separa- ally used in front of the projecting cylinders of radial engines were found to have a tion of the cooling air from the general flow. Regarding the matter of returning the negligible effect on performance. used cooling air to the general flow again, suggestions were obtained from the staff When the new N. A. C. A. cowling completely enclosing the engine was first in charge of the five-foot atmospheric wind tunnel of the laboratory, who had done tested, an exceptionally low drag was obtained, but the engine ran much too hot. some work on wing slots and boundary layer control. As a result, the new cowling The cowling was modified and retested many times before the engine cooled prop- was designed with an annular slot extending entirely around the circumference of erly, but finally the cooling was approximately as satisfactory as with the uncowled the body a short distance behind the engine. In section the slot was similar to some engine. To accomplish this, it had been necessary to enlarge the hole in the nose, wing slots in which the air passing through is directed tangentially along the sur- provide cut-outs over the magnetos, enlarge the slot, and guide the cooling air past face. Rough comparative tests on two small model fuselages were then made in the the hottest parts of the cylinders. In this connection, there is still room for improve- six-inch wind tunnel, and these indicated that a substantial reduction in drag and a ment in the effective use of the cooling air, with a possible further gain in perfor- reasonable flow of air through the nose and out of the slot would be obtained. The mance. flow of cooling air is helped by the fact that the outside air pressure at the nose is the The reduction in drag of the cabin fuselage with the final cowling, which cooled highest and that at the slot about the lowest found over the entire body, when the satisfactorily with the engine completely enclosed, was 2.6 times as great as with the body is in motion with respect to the air. best conventional type with a large spinner. A still greater difference is to be expected A program of tests was drawn up including both extremes of cowling and sev- in the case of a small fuselage or a nacelle, the tests on which are now underway. eral conventional intermediate steps, both with and without spinners and on a cabin Rough calculations show that the power required to drive an average commer- type and an open cockpit type fuselage. This program was then submitted to the cial Whirlwind engined cabin airplane at its maximum horizontal speed would be manufacturers for criticisms and suggestions, several of which were adopted. As the reduced by from 15 to 20 per cent by the use of the new N. A. C. A cowling in tests were finally made, the cooling with each form of cowling was first investigated place of the best present conventional cowling. For three-engined transports with and compared with that for the uncowled engine. Then the cowling was modified, two wing engines the decrease in power required under similar conditions would be if necessary, and retested until the cooling was approximately as satisfactory as for from 20 to 25 percent, and for machines with small open cockpit fuselages as much the entirely exposed engine. After that, tests were made on the drag and the propul- as 20 to 30 percent. sive efficiency. If the full engine power were used with the N. A. C. A. cowling, the maximum The portion of the investigation involving the cabin fuselage has been com- horizontal speed would be increased by from five to ten miles per hour for the aver- pleted, and tests are now being made with the smaller open cockpit fuselage. age cabin airplane, somewhat more for the three-engined machines, and as much as One of the first findings of the tests was the enormous amount of drag due to 20 m.p.h. for small open cockpit planes such as single-seater fighters. the uncowled engine. The drag of the bare cabin fuselage with the engine removed Inasmuch as the drag is less with the N. A. C. A. cowling, the power available for 180 Chapter 3: The Design Revolution Document 3-9(a-b) 181

climbing is greater, and the rate of climb and the ceiling will be improved. The fuel of cowling are the placing of all accessories, especially the magnetos, in the rear (this consumption will also be reduced, the amount depending on the speed of flight. If is of course now being done in many power plants), and the provision of greater the same cruising speed is maintained, as with the old cowling, the decrease in fuel distance between the plane of the cylinders and the propeller so that a better shaped consumption will be approximately proportional to the decrease in power required. nose can be had. On the other hand, if, as is more likely the case, the same engine power will be used The new N. A. C. A. cowling, being simple and smooth in form, is easily con- in order to cruise at an increased speed, the number of miles obtained per gallon will structed. The nose piece or hood used in both the wind tunnel and flight tests was be increased in proportion to the increase in speed. An increase in range will also be built into a complete ring which was inherently stiff and strong without bracing. It obtained, which is proportional to the increase in the miles flown per gallon. could be easily removed, but it was first necessary to remove the propeller. To avoid It is believed that the improved performance with the new cowling makes the this in practice, it would probably be advisable to make the nose piece in two or air-cooled radial equal to or better than the water-cooled engine in the matter of three quickly detachable sections. When the nose piece is removed the small cowl- drag except in the case of pure racing planes on which wing radiators may be used. ing over crankcase is similar to conventional types, and most parts of the engine The results of the wind tunnel tests were so promising that it was decided to requiring frequent attention can be easily reached. No difficulties of maintenance check them in flight. The Army kindly loaned an AT-5A (Curtiss Hawk with Whirl- occurred during the many hours of wind tunnel running or in flight tests. wind engine for advanced training purposes) for the tests. The new N. A. C. A In manufacture, the complete cowling would probably cost no more than a complete cowling was then adapted to the plane by the Flight Operations Section conventional cowling without spinner, except for the nose piece. Since a spinner of the laboratory, which had also constructed all of the cowling for the wind tunnel is not required, it would cost little if any more than the conventional types using investigation, and the tests were carried out by the chief test pilot, Thomas Carroll, a large spinner. The weight of the nose ring used on the AT-5A, part of which was and his two assistants, Messrs. McAvoy and Christopher. The best of the standard made of 1-16 in. thick aluminum for convenience in working, was 27 lbs. Army AT-5A planes was taken from the line and a direct comparison was made In the test cowlings, the engine exhaust was directed out of the slot by means between it and the one with the complete cowling, both being flown at the same of individual stacks on each cylinder. The conventional ring type exhaust collectors time. Each test point was checked several times by each of the three pilots. could be used behind the cylinders if desired, or if the engine exhausted at the front, The plane with the new N. A. C. A. cowling had a maximum horizontal speed the exhaust ring could be made the front part of the nose piece. This latter would at sea level of 137 m.p.h. as compared with 118 m.p.h. for the standard AT-5A, provide a very convenient means of support for the rest the nose piece. The com- both being attained with the same engine revolutions. This represents a gain of 19 plete cowling is well adapted for the use of shutters, which could be made to reduce m.p.h due to the new cowling. The reduction of engine drag is probably directly flow of cooling air over the entire engine if desired. It would seem that this would responsible for about 13 or 14 mi. of this gain, the rest being due to the decrease in improve the operation of cooled engines very appreciably in cold climates. induced drag and the increase in propeller efficiency at the higher speed. One point regarding the application of this cowling is worth mentioning. The All of the pilots reported the plane with the new cowling smoother to fly and many modifications, which were necessary before proper cooling was obtained with better in answering the controls than the standard plane. No doubt this can be the cowling, show that it must be carefully designed. It is possible that eventually attributed to the smoother airflow over the fuselage and inner portion of the tail the engine manufacturers will find it advisable to furnish the engines complete with surfaces. cowling, thus ensuring proper cooling conditions for their products. In that case the The pilots also reported that with this airplane the new cowling did not alter the exhaust system could no doubt be very neatly incorporated in the cowling. range of vision in any useful field. The complete cowling, of course, cuts off whatever In appearance, the N. A. C. A cowling is reminiscent of the hoods enclosing view can be obtained between the cylinders. Incidentally, there seems to be a great the old rotary engines of the war period. It gives the fuselage-engine combination difference of opinion regarding the usefulness of the vision which can be obtained longer and smoother lines with, even on a small fuselage such as that of the AT-5A, between the cylinders, some pilots maintaining that it is essential in the case of some are not unpleasant to the eye. planes, such as fighters, and others believing that it is never used to an appreciable In conclusion, it would seem from the test made to date that a very substantial extent. It undoubtedly depends to some extent on the contour of the engine and the increase in high speed and all-round performance can be obtained on practically all amount of space between the cylinders. If the complete cowling comes into general radial engined aircraft by the use of the new N. A. C. A complete cowling. use, as these first tests seem to warrant, there may be a tendency toward the development of more compact engines having a smaller overall diameter. Other engine developments which would improve the effectiveness of this type

Document 3-10(a-c) 183

Document 3-10(a-c)

(a) H. C. H. Townend, “Reduction of Drag of Radial Engine by Attachment of Rings of Aerofoil Section,” British Aeronautical Research Committee Research and Memoranda 1267 (1929).

(b) United Aircraft and Transport Corporation Technical Advisory Committee, Meeting Minutes, December 1929, Boeing Historical Archives, Seattle, Washington, pp. 213-256.

The NACA cowling was not the only method of reducing the drag of air-cooled engines. In 1929, Hubert C.H. Townend of the British National Physical Labora- tory also developed a low-drag engine cowling. Known as the “Townend ring,” this design also reduced drag with little degradation of engine cooling and was widely adopted in Europe and America. Because it completely enclosed the engine, many people in the aviation industry believed that the NACA cowling was inevitably detrimental to engine cooling; therefore, they intuitively favored Townend’s design. As indicated in the second document below, members of the United Aircraft and Transport Technical Advisory Committee (curiously of which Fred Weick was briefly a member) reflected this prejudice. Several Boeing aircraft, such as the P-26 “Peashooter” of the early 1930s, utilized the ring cowling. Other manufacturers employed similar structures. One company, Curtiss-Wright, developed its own form of Townend ring, which it called the “Curtiss Anti-Drag Ring,” the subject of the third document below. The history of the NACA cowling-Townend ring rivalry has yet to be written. In the beginning, neither the British National Physical Laboratory nor the American NACA appear to have been aware of the other’s cowling work. The NPL published the results of its ring research just before the NACA’s cowling reports appeared. To impress American manufacturers with the value of its cowling, the NACA placed its design into some direct competition with the Townend ring. For example, it did so with the Martin B-10 when it competed with the Boeing B-9 for a large army contract in 1932 (see Document 3-19). The overall competitive situation between the cowl and the ring fed the fire of what became a transatlantic dispute and resulted in a long series of patent suits. 184 Chapter 3: The Design Revolution Document 3-10(a-c) 185

Document 3-10(a), H. C. H. Townend, “Reduction of Drag of Radial Engine by Aerodynamics. Attachment of Rings of Aerofoil Section,” British Aeronautical Research When an aircraft is in motion, the disturbance produced in the air, together Committee Research and Memoranda 1267 (1929). with the reaction up on the machine itself and its consequent performance all depend ultimately on the nature of the flow over its surface. This “boundary flow” is AERONAUTICAL RESEARCH COMMITTEE therefore of great practical importance, and detailed investigations of its characteristics have occupied the attention of the Committee during the year. Already in the Report for the year 1929-30 previous year, Professor B. M. Jones had drawn attention, in R.&M. 1199, to the large difference existing between the drag of an aeroplane and that estimated from Brigadier-General The Rt. Hon. The Lord Thomson, ideal conditions. The coefficient of frictional resistance to the motion of a flat sur- P.C., C.B.E., D.S.O., p.s.c., Secretary of State for Air. face through the air is known from experiment to depend on the Reynolds number at which the experiment is made. In the ideal conditions it should be possible to June, 1930 calculate the resistance to an aeroplane from a knowledge of its surface area and of this coefficient of friction, on the assumption that its value does not depend on the My Lord, curvature of the surface. An enquiry has been started on the curvature of the surface. The Aeronautical Research Committee beg to submit their report for the year An enquiry has been started to investigate how far this assumption is justified. 1929-30. The frictional coefficient depends also on the nature of the flow over the sur- The Committee wish to draw attention to the stead expansion of activities in face. This may be “laminar,” in which case the air moves in smooth curves follow- aeronautics throughout the world and to the consequent increase in the number of ing more or less the shape of the surface, or it may be “turbulent” when a series of problems awaiting solution which have been brought to their attention in connec- eddies is formed. For turbulent flow the coefficient considerably exceeds that found tion with developments within the British Empire. when the flow is laminar. In experiments in a wind tunnel both kinds of flow are The speed and size of aircraft continually increase and the number of uses to observed; the flow over the forward part of a good streamline model is laminar, over which aircraft are put grows steadily. New flying boats and aeroplanes, fitted with the rear it is generally turbulent. As an aid to exploration near the surface of a body, a number of engines, are projected. The speed of aircraft, especially for service and an instrument has been designed which depends on the rate of cooling of a fine racing purposes, has increased greatly; the race for the Schneider Trophy in 1929 and electrically heated wire and responds readily to fluctuations in the airflow. When the speed record made later over the three kilometer course by officers of the Royal placed near the surface of a model, this instrument can be arranged to indicate by Air Force indicate the extent of this advance. The new airships, R.100 and R.101, audible means the nature of the flow and the points, if any, at which a change from have completed their trials during the year and as a consequence of experience with laminar to turbulent flow takes place. them, the question of larger ships, bringing with it new problems both of design In some earlier investigations of the same problem, a small pitot tube was moved and performance, has naturally come forward. Research requirements demand not near to and parallel to the surface of the model. Some difficulties which attended only the solution of these new problems but also experiments on larger models at this method have been overcome and the hot wire and pitot tube methods are now higher values of the Reynolds number, i.e., at higher speeds or in air under pres- in good agreement. In addition to these two methods, the smoke trail from titanium sure; small models do not always reproduce the actual aircraft in sufficient detail. tetrachloride, a chemical very suitable for this purpose, has made details of the flow For work at higher values of the Reynolds number a compressed air tunnel is under visible. The first experiments with the chemical have given a good indication of the construction at the National Physical Laboratory, while for models requiring engine places on the models where the changes from laminar to turbulent motion occur details the erection at the Royal Aircraft Establishment of a large wind tunnel with but the technique of its use must be regarded as still in its infancy. an air jet 24 ft. in diameter has been recommended. In addition, a scheme has been In addition to the study of the flow past a streamline body, a Joukowski type of put forward for replacing the obsolescent 7 ft. tunnel No. 1 at the N.P.L., by two aerofoil has been selected for experiments on resistance, for pressure plotting and for open jet tunnels of 8 ft. diameter housed in the existing building. This moderniza- measurements of pitot head at distances as small as 2 or 3 thousandths of an inch tion will materially accelerate the progress of work in the N.P.L. program. from the surface. A report on this work is being prepared and will be published at an early date; it has a special interest since the form employed in this type of aerofoil is one for which the characteristics of the flow can be determined theoretically, so that comparison between theory and experiment will be possible. 186 Chapter 3: The Design Revolution Document 3-10(a-c) 187

The flow over airship bodies and aerofoils is generally of a stable nature, whereas A cognate research on airscrew body interference, for both tractor and pusher the nature of the flow past a circular cylinder is known to vary rapidly at certain screws using ideally streamline bodies, is well advanced. The experiments on the speeds. As another means therefore of attack on “critical flow,” the experimental tractor screws are complete and those on pusher screws are in hand. This will provide conditions for a cylinder have been changed by setting up artificial turbulence in a basis for further interference experiments upon airscrew-body-wing combinations. the airstream and by the alteration of surface roughness. These changes produced Concurrently with the general investigations on interference and, indeed, aris- orderly responses in the characteristics of the flow and it appears that even in the ing out of them, a valuable device known as the Townend Ring has been developed, range of the Reynolds number in which there is a large change in the resistance coef- whose use leads to a substantial reduction of the drag of aeroplane bodies fitted with ficient these characteristics are quite definite. radial engines (see Illustration No. 2). The effect of the ring is to direct the air along The above experiments were all made at normal wind tunnel speeds. In other the body and lessen the amount of turbulence; the higher the lift coefficient of the work carried out at very high speeds, pressure distribution has been measured round cross section of the ring the more efficient is the Townend Ring in decreasing total a Joukowski section. Up to a speed of half the velocity of sound the force coefficients resistance to forward motion. Moreover, the cooling of the engine is not adversely vary by only a small amount from those found at low speeds but there is an appre- affected. An account of the experimental work on this Ring has been issued in ciable divergence, when the speed reaches about 0.6 times the velocity of sound. the Reports and Memoranda Series; since its publication, the results obtained with On other wing sections closer agreement with results obtained in a low speed wind models have been confirmed by a number of full scale trials. In one case the top tunnel was obtained as regards lift, but as regards drag there was a marked difference speed of a certain aeroplane was increased by eight miles per hour. even at velocities only half that of sound. The evolution of the Townend Ring affords an example of the valuable practi- A theoretical study allowing for the compressibility of air has also been made of cal results that may follow from simple experiments originally intended to throw the conditions under which the flow at high speeds past curved surfaces may exceed light on fundamental points. The success achieved in this case has encouraged the the velocity of sound close to a convex surface, while remaining below that speed Committee to continue their policy of investigating first the simplest cases of body at some distance from the surface. There appears in the mathematical solution pre- and wing combinations in endeavoring to establish the main principles on which sented to be no indication of any discontinuity of the flow; moreover, speeds just interference depends. greater than that of sound are attained without the formation of compressibility waves. This work explains why the experiments in the electrolytic tank at Cam- Performance. bridge mentioned in last year’s report, were not successful at speeds above a certain The variation of the range of an aeroplane with speed and height is of con- value. Some confirmation of this theoretical work has been obtained in the small siderable practical interest. Experiments made at the A. & A.E.E., Martlesham, high speed jet at the N.P.L., but the matter is not yet completely cleared up. showed that the maximum range of a particular aeroplane increased substantially with height whereas theory indicates that a change in altitude should have little or Interference. no effect. While much time has thus been spent on studying flow near the surface, the effects due to the interference of one part of an aircraft on another have not been REDUCTION OF DRAG OF RADIAL ENGINES BY THE neglected. Attention had previously been drawn to the fact that aeroplane bodies and wings differing only in the method in which the wings were attached, gave ATTACHMENT OF RINGS OF AEROFOIL SECTION, INCLUDING widely different drags. The early experiments to elucidate this matter, described in INTERFERENCE EXPERIMENTS OF AN ALLIED NATURE, WITH last year’s report, were purposely made upon a streamline body thicker and shorter SOME FURTHER APPLICATIONS. than the average aeroplane body, with the object of accentuating the effects to be measured. Selected experiments have been repeated upon another body of propor- By H. C. H. Townend, B.Sc. tions more nearly similar to those of the bodies of aircraft and the researches are being extended to include an analysis of the separate forces on wings and bodies. Reports and Memoranda No. 1267. Moreover, a series of experiments on the interference of an engine nacelle and a (Ae. 413.) wing has been commenced in the largest N.P.L. (Duplex) wind tunnel; an airscrew (Contents see pp. 538, 539). and an aeroplane body will be added later. The effect of changes in the relative position of the body and wing for the cantilever monoplane type of construction is July 1929. being investigated at the R.A.E. 188 Chapter 3: The Design Revolution Document 3-10(a-c) 189

Summary. Introductory. Some experiments are described in which large, and Further developments. Full scale tests are in hand at the R.A.E. in which mea- frequently negative, interference effects are found to be produced by certain objects surements are to be made of the effect of two typical rings on the performance of a of streamline or aerofoil form upon the drag of (1) a model of airship form (U.721), Bristol Bulldog aeroplane (Jupiter engine). (2) model aeroplane bodies having radial engines in the nose, and (3) models in The rings selected for test on this machine are the full scale equivalents of Ring which turbulence is produced by grooves or sharp edges. The evidence provided by J and polygonal Ring P3 shown in Fig. 25 of this report with a slight modification the tests shows that the effects produced in cases (2) and (3) are due, not to shielding to the angle between the chord and the body axis in the latter case. or fairing of the obstructions, but to the addition of small aerofoils to the models in The influence on the cooling of the Engine is also to be measured. such a way as to control the airflow in the neighborhood of the obstructions, chiefly The experiments described in this report were commenced in December, 1927. by governing its local direction. The results include those given in the two earlier reports cited. The report is divided In case (1) there were no obstructions on the body, and the results are con- into two parts: cerned simply with the interferences between the body (of airship form) and rings Part I. Interference on the drag of a model airship form (U. 721) caused by of streamline section surrounding it or of streamline struts and aerofoils near it. The the presence near the nose of struts and aerofoils, and of rings of streamline section effects on the body were very marked, its drag falling to zero when surrounded by coaxial with the body, in various positions relative to the nose. a ring of as much as three times its own maximum diameter situated near the plane Part II. The reduction of the drag of aeroplane bodies fitted with radial engines of the nose. For struts or rings in closer proximity to the surface, the body drag in the nose by the addition of rings of aerofoil or streamline section. Some miscel- was negative in direction and three or four times its normal magnitude. There is at present no satisfactory explanation to be offered for some of these latter results, laneous results are also given in this Part. through it is hoped that work at present proceeding or projected will help to throw The two parts will be treated separately but the results of Part II will be pre- some light on them. In one case of this kind the combined drag of body and ring, sented in greater detail in view of their immediate application to the problem of the when suitably disposed to one another, was less than the sum of the drags of each in drag of the air-cooled radial engine. free air although the presence of the body might have been expected to increase the ring drag above its minimum value, by making the direction of the local airstream PART 1 oblique to the ring section. INTERFERENCE OF STRUTS AND RINGS OF AEROFOIL SECTION ON A MODEL AIRSHIP FORM (U.721). Applications. The chief result of the experiments has been the development of a method of reducing the drag of radial engines, which consists in placing a ring of 1. Introduction.—Origin.—In the course of a test connected with a proposed aerofoil section round the nose in front of the engine and partly overlapping it. The investigation into the problem of airscrew body interference it was observed that the aerofoil section adopted for the ring may vary widely from a cambered plate to a drag of a streamline body mounted in a wind tunnel was very considerably affected thick symmetrical section such as could be used for an exhaust pipe or silencer. by the presence of a ring of streamline section surrounding it but independently The magnitude of the reduction obtained with a given ring increases with the supported from the tunnel floor. Although the diameter of the ring with which this number of cylinders in the engine, at least up to nine. With 14 cylinders in two observation was made was three times the maximum diameter of the body, and its rows the reduction is practically the same as with 9. In the best case tested (with 9 cross section relatively small, the observed effects were large, the changes in drag cylinders) a reduction of drag was obtained equal to 60 percent of that of body and being much greater than the total undisturbed drag. In view of this it was consid- engine only (i.e., R/R = 0.40). o ered worthwhile to conduct a few experiments aside from the original program to Full scale tests indicate that the cooling of the engine is usually unimpaired investigate the effect, particularly as the drag was in some circumstances actually and in some cases improved. Considerable silencing may also be effected when the ring is used as an exhaust collector. The ceiling may be increased (see 13). The ring reduced—in one case to zero. With this object the following tests were made. interferes but little with the accessibility of the engine, and involves no modification in the body whatever; in fact, it is probably that in most cases the spinner and 2. Description of Tests. (a) Tests with Rings near a streamline body.—The body in some cases even the crank case cowling may be discarded without appreciably upon which the experiments were made was a model of the airship form U.721 of affecting the drag. maximum diameter 6 ½ in. and fineness ratio 4.622, having the low drag coefficient The majority of these full scale tests have not been strict comparisons with the of model tests reported here, although based on them; exceptions to this statement will be noted in the text. R/ρV2(vol)2/3 = 0.007 approx., (R/ρV2d2 = 0.0117) 190 Chapter 3: The Design Revolution Document 3-10(a-c) 191

The body was supported on wires from the roof of the (4 ft.) tunnel and from (a) the diameter of the ring; the main bottom balance in the usual way; the ring was mounted on a spindle (b) the orientation of the section of the ring (i.e., its incidence relative to the attached to a large flat metal plate 3/16 in. thick which could be traversed along the local streamlines); tunnel floor. The axes of tunnel, body and ring were coincident. (c) the size of the section. In the initial experiment the diameter of the ring was 18 in. at its leading and However, to carry out such tests on rings would have been quite out of the trailing edges, and the chord of its (symmetrical) section 2 in. long. It was expected question on account of the large number that would have been required, so it was that the interference of so large a ring would be small; in fact the original tests was decided to open out the ring, as it were, into a straight strut, which might logically made with the object of verifying that it would be so. To accentuate its interference, be expected to give results of the same general nature as those obtained with the therefore, the ring was made with a thick section only roughly faired (shown at γ, ring. A few rough preliminary tests showed that this was the case, although it was Fig. 1). evident that when close to the body, the rapid change in the incidence of the strut The results obtained with the first ring showed an unexpectedly large interfer- along its length due to the great curvature of the streamlines near the body, as well ence effect on the drag, a noticeable feature being that when placed in the neigh- as the rapid change in the distance of an element from the body surface, would borhood of the nose of the body the drag was reduced. The ring was then reversed, preclude a strict comparison. that is, placed trailing edge foremost, to make the section somewhat “worse”. As The substitution of struts for rings enabled (a) and (b) above to be varied with this appeared to make the effect of the ring everywhere less, an attempt was made ease whilst (c) presented no difficulty, and the great resulting experimental simpli- to improve the section by fairing it with plasticine, on the assumption that by so fication was considered to justify the strut tests apart from any interest they might doing the effect would be correspondingly increased. This was found to be the case; have in themselves. A description of the strut tests is given in section 2.3 (b). so much so in fact that the minimum drag (which occurred when the nose of the The strut tests suggested that the lowest drags would be obtained when the body and the leading edge of the ring were coplanar) was reduced to zero (Fig. 1). chord of the ring section (all the sections tested were symmetrical) was roughly par- As these changes were consequent solely upon a change in the section it was decided allel to the local direction of the streamlines. The nose of U.721 is of almost exactly to test a ring having the section of a good strut of fineness ratio 2.67, but of only the same shape as a 4 to 1 spheroid, and the streamlines and their directions near the ¾ in. thickness (referred to as section α, see Fig. 1). The effect of this ring was nose, were calculated from the appropriate formulae for a spheroid given in Lamb’s somewhat smaller, no doubt due to its reduced thickness, but it exhibited the same Hydrodynamics. The results are shown in Table 13 and Fig. 11. general characteristics. For the sake of completeness three further rings were tested Two rings (A and B) of section α (Fig. 1) were then made with [the] chord of decreasing cross sectional area, of the following sections: lines along the local streamlines when in position x = 0, [as] defined previously. (a) rectangular, 1/8 in. X 1½ in. These are shown in Fig. 2. The drag of [the] wings themselves was also measured by (b )circular, of diameter 5/16 in. mounting them on a spindle in the bottom balance and fixing the body on a stand (c) circular, of diameter 1/8 in. which could be traversed along the tunnel. The forces on rings and body separately for different relative positions are shown in Fig. 3. They are relatively unimportant The results are shown in Fig. 1 where the drag of the body (in lb. at 60 ft./sec.) however, as a result of adding the appropriate curves for body and ring shows that is plotted against the axial position, x (inches) of the ring relative to the body (x = 0 the overall drag is always more than twice that of the body only, but is, generally, when leading edge of the ring is in plane of nose of body, and is positive when ring is considerably less than that of the ring alone, which is perhaps not surprising in view entirely aft of this plane). It will be seen that the last 3 rings produce no perceptible of the angle of the chord line for these rings. More interest attaches to the dotted reduction in drag anywhere, but that when amidships even the ring of 1/8 in. wire curves which show the drag of the ring when seven radial cylinders are attached to still has an appreciable effect. the body. The modification in the case of Ring A is very marked; this feature is of The effects of these large rings (18 in. diameter), were observed for all axial posi- practical importance and will be referred to in Part II. tions of the ring relative to the nose of the body. Subsequent smaller rings were only tested in the neighborhood of the nose, where a reduction in the body drag occurs. 2.1 Effect of Rings on a Body having an annular excrescence.—After the above In view of the magnitude of the effects observed with rings of such large diam- tests were made a ring of rubber tubing (0.2 in. outside diameter) was nailed to eter, particularly when the sections were of streamline form, it seemed possible that the body in contact with its surface in the plane x = 3.8 in. The effect of the rings interesting results might be obtained, and possible some insight into the cause of A and B on the drag of the body so modified was then measured. The results are the phenomenon by varying not reproduced but may be expressed generally by remarking that the shape of the body drag curve (Fig. 3) with Ring B present is almost unaltered but the curve is 192 Chapter 3: The Design Revolution Document 3-10(a-c) 193

raised bodily by about 0.055 lb. at 60 ft./sec. while the drag of the body with ring absent is raised by about 0.075 lb. The effect of ring B is therefore somewhat greater when the excrescence is attached than when it is not. Ring A on the other hand has a smaller effect when the excrescence is attached, the curve being raised to a greater extent than the drag of the body only and being somewhat flattened. The overall drag was not measured.

2.2 Results obtained with Rings.—With all rings tested the overall drag is greater than that of the body only, but is frequently less than that of the ring alone, and is almost always less (in the most favorable position) than the sum of the separate drags. This latter fact would be expected in the case of the smaller rings on account of the angle at which the chord line is set. There would also be a local change of pressure between the ring and the body giving rise to axial forces superimposed on the drag, properly so called, and in consequence the forces on ring and body taken separately have little significance. In any case the drag of the ring with the body absent would be high on account of the inclination of the chord to the free stream. It is, however, somewhat surprising that the same effect should also be observed in the case of the 18 in. diameter ring (section α), for in this case the section is already in the position of minimum drag when the body is absent, since the chord line is then parallel to the wind. It may be noticed that here the drag of each is reduced simultaneously, implying a reduction in the total turbulence apart from any pressure reactions. The introduction of the body would slightly increase the air speed at the ring, and so tend to increase its drag. No explanation of this effect is offered unless it is to be found in the fact that the ring section is somewhat critical or had its minimum drag at a finite angle of incidence. It would be worthwhile making some further search for the cause of this effect. 2.3 Description of Tests. (b) Tests with Struts and Aerofoils.—The struts under test were mounted on vertical spindles roughly coaxial with their centers of pressure, and were disposed in pairs, symmetrically on either side of the nose of the body. They were capable of being yawed about the axes of the spindles, which passed through holes in the floor of the tunnel. The upper ends of the struts were braced by wires to the walls of the tunnel. Attention was mainly concentrated on those positions of the struts which produced a reduction of body drag. The forces of the struts were not measured. In Fig. 4 the sections and positions of the struts when at zero yaw, are shown in relation to the nose of U.721, together with the axis of yaw in each case. It was found that the change in drag produced by two struts was almost exactly double that produced by one alone (see H and 2H, Figs. 4 and 5) when each strut was only 2 in. from the body axis; most of the tests were therefore made with one strut only. The strut of section α (A to F, Fig. 4) was 3 ft. long; all the rest were 11 in. Aerofoils G, M and N were of Gottingen 429 section; aerofoils H, K and L were (approximately) R.A.F. 30 section. 194 Chapter 3: The Design Revolution Document 3-10(a-c) 195

2.4 Results obtained with Struts.—The results of the foregoing strut experiments effect it was recalled that many aeroplanes powered with radial engines are provided are shown in Figs. 5 (a) and (b), where the drag of the body (lb. at 60 ft./sec.) is with annular exhaust pipes running round the nose of the machine just in front plotted against angle of yaw of strut relative to the tunnel axis. The angle of yaw is of or behind the cylinder heads. In normal circumstances, such exhaust rings are positive when the leading edge of the strut is nearer the axis of the body than the made of circular or elliptical cross section and usually add quite considerably to the trailing edge. already high drag of the radial engine. Even when made of streamline section they It was thought unnecessary to tabulate the large number of readings taken, as increase the drag as normally fitted. the curves were well defined including the critical region near—5° yaw. There was however, in this possible application one serious point of difference The following points may be noticed: from the previous experiments which it was thought might vitiate the application of (1) the reduction in drag becomes less as the aerofoil is moved downstream the results to the case when the radial engine was present, and that was the influence from the plane of the nose. of the cylinders on the air flow; but as the ring had been found to be still effective (2) Large reductions may occur even when the aerofoil is some distance from when an annular excrescence was attached to the nose of the streamline body in the the body laterally, see A, Fig. 4. position which would be occupied by a radial engine (see 2.1) it was considered (3) Aerofoils C and G, though of widely different sections and slightly different probable that this case also would be amenable to similar treatment, by a suitable lengths of chord, produce, in the same position, nearly the same effect. modification of an existing exhaust pipe or by the addition of such a pipe when not In addition to the above a few tests were made with a pair of aerofoils of Got- already fitted. With this objective in view it was decided to make a few preliminary tingen 429 section 11 in. long similar to M, Fig. 4, mounted on a stand in such a tests. way that they could be set at any angles of yaw and could be traversed together along The results of these tests were very encouraging, and showed at once that appre- the tunnel over a [range] of x near the nose. ciable reductions in drag were obtainable. They also disclosed the mechanism by The axes of yaw, which were situated at ¼ chord from the leading edge of the which the ring produced its effect (in the case of the radial engine) and this informa- aerofoils, were spaced 3.35 in. apart transversely. The drag of the body was mea- tion was used to design other rings giving much better results. sured for angles of yaw of both aerofoils of 15°, 25°, 35° and 45° over a range of x In order to render the description of the tests more intelligible it may be as well which included minimum drag. at this point before describing them to anticipate the results, for the purpose of The results are shown plotted in Fig. 6. The chief points of interest are as follows: giving an indication of the action of the ring. (1) The reduction in drag continues to increase up to quite large angles of yaw. 3.1 Function of the Ring.—The distribution of streamlines around the nose of (2) For a given angle of yaw, the reduction in drag continues to increase as x a streamline body, when it is not fitted with a radial engine is such that the flow increases until the aerofoils are quite close to the surface of the body. Minimum continually converges towards the surface of the body as it proceeds from the place drag occurs when the minimum clearance is of the order of 3/16 in. On closer of the nose to that of the maximum cross section. When the cylinders of a radial approach the drag becomes positive before actual contact occurs between aerofoils engine are placed in their usual position in the nose they disturb this ideal condi- and body. There is therefore a stable relative position, in which the aerofoils, though tion and produce a violent deflection of the airstream away from the surface, this not touching the body are “towing” it against the wind. being so not only in the case of air which directly strikes the cylinders, but also of the air which passes between them. This effect can be observed by discharging a jet PART II of smoke in front of the gap between two cylinders. The Reduction of the Drag of Radial Engines If by some means this divergence of the streamlines could be obviated it might 3. Introduction.—In paragraph 2.2 (Part 1) tests were referred to in which the be supposed that the inevitable disturbance produced by the cylinders themselves interference between a ring and body was negative, i.e., the combined drag was would be prevented from expanding to form a large turbulent wake, with a con- lower than the sum of the separate drags. This effect is not uncommon when the sequent reduction in drag. Now the attachment of a streamline ring is a method interfering bodies have a high drag, although in such cases the explanation is usu- of effecting this, since it consists in placing immediately in front of the cylinders a ally that one of the bodies tends to shield the other. In fact, ordinary fairing is an curved aerofoil disposed in such a way with respect to the cylinders and the local extreme example of this. In the case of the 18 in. diameter ring of section α already direction of the flow as modified by them that the aerofoil section, working at a referred to (2.2) the two interfering objects not only did not shield one another but fairly high lift coefficient, secures control of the air before it is thrown outwards were low drag forms to start with, and yet their mutual interference was negative. by the cylinders, by exerting a “downwash” directed towards the body axis which In seeking for some practical example in which advantage might be taken of this prevents the disturbance caused by the engine from expanding radially to produce 196 Chapter 3: The Design Revolution Document 3-10(a-c) 197

a large wake. From smoke observations (visual) with and without a ring it was evi- area as that of a 0.58 in. diameter circle. The clearance between the segments and dent that the ring reduced the radial thickness of the wake behind the cylinders by the body was about ¼ in. to 3/8 in. something like 40 percent. Two complete rings (C and D, Fig. 2) based on the above tests were then made Some further visual experiments were made on a very small (dimensional) and attached to the body with 7 radial cylinders in the same plane as before. The model in a water channel, in which the changes in flow were made visible by a jet of results agreed closely with expectations based on the segment tests, as did also tests dye discharged in front of the model. Photographs of the flow are shown in Fig. 29. made with rings of 0.58 in. diameter circular cross-section. In each photograph the jet is in the same position. In the first (a) the flow past the Several rings (A, B, C and D) were then tested for drag in the presence of the body only is shown. In (b) two exposures were made on the same plate, the first a body (with 7 cylinders), and also the body and cylinders in the presence of the rings. repetition of (a) and the second with the cylinders in position; the line of ink shows The results (Table 7) are shown in Figs. 7, 8 and 9. The curves for ring and body the flow between two cylinders. In (c) there are also two exposures on one plate; are added graphically, the sum being shown by the dotted curves, while the points the first is a repeat of (b) with cylinders, and the second is the flow with an aerofoil with tails are overall observations made with the rings actually attached to the body. attached in addition, to represent a ring. The drag of the rings has been corrected for spindle effect which was measured by means of a dummy and in which there may be slight uncertainty. The important (overall) observations are not subject to this uncertainty. They have been corrected 4. Description of Preliminary Experiments.—In searching for a typical aeroplane throughout for static pressure drop, and drag of wires and spindle, but errors in which should have a body similar in shape to the model (U.721) already used for the these corrections do not affect the comparisons. The forces have not been reduced

experiments of Part I, an inspection was made of numerous photographs appear- to coefficients but are shown in lb. at 60 ft. per second wind speed. Scales of R/R0

ing in “Aviation.” It was found that the U.S. Navy “Apache” biplane fitted with a (where R is drag of body and cylinders with ring attached, and R0 = drag of body “Wasp” radial engine had a fuselage the shape of which in side elevation was almost and cylinders only) are drawn on the right hand sides of the figures. identical with that of U.721 up to the maximum diameter. From a few rough mea- In addition to the ring of 0.58 in. diameter circular section, rings of circular surements of one such photograph seven model cylinders were made and attached section were also tested having the same frontal area as that of rings C and D, i.e., to U.721 in the appropriate position. The cylinders were made to fit the surface of 0.3 in. radial depth, in the same relative positions as C and D respectively. These of the model closely and each was held on by one central pin so that it could be results are plotted at x = −3 and at x = 0.1 in Fig. 7. removed and replaced easily. The shape of the cylinders and their situation in the Some further tests subsequently made on this model with other rings are nose are shown in Figs. 2, 13a and 14. They were very simple in outline and were described later together with the results of more extended experiments. fitted with two dummy cylindrical valves apiece. 5. Description of Extended Experiments.—As the initial experiments were suf- An initial attempt to discover the most suitable angle for the ring section rela- ficiently promising to warrant a more thorough investigation, experiments were tive to the body axis under the new conditions was made by using 4 cylinders only, instituted on larger models in which the details of the engine cylinders were repre- a pair on each side of the body, each pair being spaced as though forming part of a sented more accurately. radial engine. Behind each pair was fixed a segment of a ring of symmetrical stream- Since the circulation around the aerofoil section of the ring associated with line section (Section β, Fig. 2). Each segment was attached by two pins (diameter the lift would produce a velocity tending to oppose the flow between the cylinders 0.034 in.) to the body, so that by bending the pins the angle of the chord to the which might be important in connection with the cooling, a few observations of the body axis could be varied to some extent. The segments were set to various angles change of air speed between the cylinders were made with a small pitot tube with and the drag was observed and plotted. It was found that there was a minimum various rings in position. which was appreciably lower than the drag of the body and cylinders alone. When One model was fitted with an airscrew, which enabled measurements to be placed just behind the cylinders the best angle of the chord was found to be 3° and made of changes in net thrust, torque and efficiency due to the ring. when placed in front the best angle was 14°. These angles apply, of course, only to The models tested were: the axial and radial positions of the ring section shown as C and D, Fig. 2 (1)U.721 model described already (see A, Fig. 13). Similar tests were made with segments of circular cross-section to imitate the (2)¼ scale model of Short Crusader Seaplane (9-cylinder engine) with and more usual type of exhaust pipe. From photographs, the appropriate diameter was without wings and floats (D), Fig. 13 and Fig. 28. judged to be about 0.58 in. These tests showed that the drag was very considerably (3)¼ scale model (modified) of body of Bristol Bulldog with 9-cylinder engine, increased. The streamline section β was designed to have the same cross sectional with and without airscrew (C), Fig. 13 and Fig. 27 (a) and (b). 198 Chapter 3: The Design Revolution Document 3-10(a-c) 199

(4)1/5 scale model of Siskin aeroplane body with Jaguar engine (14 cylinders in two cases—for equal increments of ψ in the case of 2 dimensions, and in 3 dimen- 2 rows) (E), Fig. 13 and Fig. 27 (c). sions for such values of ψ as make streamlines intersect those for 2 dimensions at (5)1/6 scale model of body with monoplane wings and Lynx engine (7 cylin- points in an axial plane 2 in. behind the nose. The difference in slope between the ders) (B), Fig. 13. 2 and 3-dimensional cases is given in Table 13 for various positions near the nose. In the important region the slope in 2 dimensions is about 6° steeper than it is in 3 5.1 Two-dimensional Experiments.—Before commencing the above tests, it was dimensions. An inspection of the observations given in the tables will show that in deemed worthwhile to obtain, if possible, some general qualitative information cases where the ring and aerofoils correspond closely in other respects the value of θ about the effect of varying the chord and section of the ring and the angle of the for minimum drag in 2 dimensions is about 8 or 9 degrees greater than in 3 dimen- chord to the body axis, by an adaptation of the “strut” method described in Part I sions. A rigorous test would have required 3 or 4 rings differing only in the angle θ, (b), in which not only the ring but the body also would be reduced to 2 dimensions. and this was not considered worthwhile. For this purpose a model was constructed of exactly the same sectional shape and size as that of U.721, Fig. 13A, but of approximately 2-dimensional form. A sketch 5.3 Drag in 2 dimensions.—The drag of the 2-dimensional model with and of this, as mounted in a 4-ft. tunnel is shown in Fig. 10. The “span” of that part on without cylinders was of course very much higher than that of U.721, and there-

which the drag was measured was 10 in., extended by a further 10 in. at each end fore the value of R/R0 has very little significance in this case. It was found however by dummies attached to the roof and floor of the tunnel and separated from the that the change in drag (∆R) obtained by the addition of a pair of aerofoils at their ends of the model proper by air gaps [1/14] in. wide. It was considered unnecessary best position corresponded somewhat closely with the change produced in U.721 to carry the dummies as far as the roof and floor; the upper dummy was slung on by the addition of the corresponding ring, the agreement being better than might radius rods and was capable of being swung out of position to allow of easy angular have been anticipated in view of the inevitable dissimilarity of the two models. This adjustment of the aerofoils by which the ring was represented. These aerofoils were encouraged some confidence in applying conclusions drawn from 2-dimenstional attached to the model by two flat brass strips provided with a row of holes ¼ in. tests to rings with respect to the effects of chord, section, variation in the angle θ apart through which screws passed into the ends of the aerofoils. The angle of an and to a lesser degree the value of θ. aerofoil was therefore adjustable about an axis whose position was itself adjustable A few figures bearing on the above remarks are collected in the following table: by varying the position of the brass strips in the body and by selecting suitable holes in the strips. The cylinders were represented by the same simple forms as were used on TABLE U.721. They were fixed in two rows of four cylinders each, one row on each side of the body in the same position relative to the nose as for U.721. They were separated Comparison of Changes in Drag (∆R) due to Rings on U.721 with Analogous Tests laterally by a distance equal to their mean circumferential separation in the U.721 in 2 Dimensions. case with 7 cylinders. Center lines of cylinders 4.0 in. behind nose. In order to preserve the interference between the cylinders each row was V = 60 f/s. extended beyond the air gaps by two cylinders at either side and at each end, fixed to the dummies. The aerofoils however were not extended in this manner. The 2-dimensional analogue of Ring A was tested to determine whether such tests would bear out the results already obtained with this ring on U.721. The result 5.2 Relation between Tests in 2 and 3 Dimensions.—It was realized that the best ∆R = 0.119 was considered to be sufficiently close to the value ∆R = 0.140 obtained angles for the aerofoil sections would not be the same in the 2 and 3-dimensional with U.721 to justify the method; other sections were therefore tested in 2 dimen- cases, and some idea of the difference was sought by calculating the slopes of the sions and from the results Rings E and G were designed. The subsequent tests with streamlines at points in the vicinity of the aerofoils for each case on the assumption these rings bore out the 2-dimensional results quite closely. that the flow near the nose of the body would not be seriously modified by the tail With regard to the angle θ later tests suggest that this should be somewhat being tapered instead of elliptical. The formulae used were derived from those given smaller (by 3 or 4 degrees) than that obtained by applying to the angle observed in in Lamb’s Hydrodynamics relating to the stream functions ψ for the elliptical cylin- 2 dimensions the theoretical correction previously referred to. der and ellipsoid respectively. A few further tests were made in 2 dimensions with 2 rows of cylinders, one To illustrate the change in angle the streamlines are shown in Fig. 11 for the behind the other, to imitate the “Jaguar” type of engine. In this case the front row 200 Chapter 3: The Design Revolution Document 3-10(a-c) 201

consisted of the same 4 cylinders per side as were used before in the same positions, The drag in this condition is taken as standard (= R0). (For body and engine only it but in addition 3 cylinders per side were added behind the spaces between cylinders is 7 percent less than that of the shape shown dotted.) in the front rows. Four cylinders only were attached to the dummies, one opposite Some previous experiments had been made on this model at the R.A.E. with each end of the front rows of cylinders on the model. helmets fitted to the individual cylinders which were designed to give sufficient Similar results were obtained with this arrangement as with a single row of cooling (see Fig. 13B). It was therefore possible to compare the relative merits of cylinders. helmets and rings for the case of 7 cylinders as well as for 9 (i.e., on the Crusader).

6. Tests of Aeroplane Models. (See Fig. 13.)—6.1. Crusader (Scale).—After the The tests made were: preliminary work on the U.721 model already described, more detailed experi- (1)Body and engine only * ments, the results of which will be discussed together later, were made on scale (2)Body and engine and wings * models of bodies of existing aeroplanes (Fig. 13) in which the details of the engine (3)Same as 1 and 2 but with helmets instead of rings. were accurately represented. The first such model used was the ¼ scale model of the Short “Crusader” high * with and without rings. speed seaplane fitted with a 9 cylinder radial engine (“Mercury”), a model which had been previously tested in connection with the Schneider Trophy Seaplane Race. 6.3 Bristol Bulldog (modified), ¼ Scale.—The body of this model was a solid of A photograph of this model is shown in Fig. 28 and a sketch in Fig. 13(D). The nose revolution in which the shape from the nose to just behind the engine was the same is shown on a larger scale in Fig. 15. The following tests were made on this model: as the Bristol Bulldog. The outline of the rest of the body was a rough mean between (1)Body and engine only, with normal cylinder fairings, cockpit, and pilot’s the outlines in plan and elevation of the Bristol Bulldog. fairings. The engine was represented by the same cylinders as used in the Crusader, but (2)Complete machine with wings, floats and tail unit, with normal fairings. the Mercury valve rockers were removed and their places taken by dummy rocker (3)Tests similar to (1) and (2) but with the individual cylinders fitted with hats of the type used on the Jupiter engine. The cylinders were so adjusted radially dummy helmets. See Fig. 13D. that the outside diameter was equal to that of the Jupiter. (4)All three tests were made with and without various rings, and in addition Fig. 15 shows the engine in position on the Bulldog model. Superimposed is a to measurements of forces, a few observations were made of velocity between the sketch of the nose of the Crusader with the position of the Mercury valve rockers cylinders. shown dotted. A visual study by means of smoke was also made of the effect produced by the 6.4 Airscrew Tests.—In the case of the Bulldog, tests were made with the air- ring on the wake behind the cylinders. screw running. The screw used was a 2-blader which happened to be available and

The results (R/R0) obtained in test (1) with Ring A, agreed closely with the cor- of approximately the correct diameter. The pitch was somewhat low (geometrical responding results for U.721 with 9 cylinders. In fact it became clear as the experi- P/D approximately equal to 0.66), but this was considered unimportant as it was ments proceeded that the magnitude of the reduction in the drag obtainable was only desired to ascertain whether the screw would alter, to any appreciable extent, not sensitive to the exact shape of the body, although the actual design of the ring the character of the results obtained with various rings. for any particular case was dependent on the shape of the nose and disposition of The screw tests consisted of the measurement of net thrust and torque. The the engine in relation to it. apparatus used, shown in Fig. 16, was similar to the standard arrangement so far as thrust was concerned, which was measured on a bottom balance. The torque bal- 6.2 Aeroplane Body with “Lynx” Engine (1/6 Scale).—This was a model of an ance was somewhat modified however. The motor was carried at the rear on a cross aeroplane body of circular cross-section fitted with monoplane wings of 5:1 aspect beam A fixed to the body. At the front it was hung directly from the torque balance ratio springing from the center of the body. A detailed drawing of the nose and B on the roof by means of 2 parallel stranded cables C, attached to the torque arms engine of this model is shown in Fig. 14, superimposed upon a sketch of U.721 and on the motor casing. The torque arms were completely enclosed within the body. cylinders, with which the general shape and size corresponded fairly closely. The To prevent lateral motion of the motor relative to the body, the nose of the motor nose cap (A, Fig. 14) was removable, the normal outline being shown by the dotted was constrained between 2 vertical strips of metal (not shown) attached to the inside lines, but for the purposes of the present tests it was considered fairer to substitute of the nose of the body. A slight clearance was allowed between the strips and the the nose cap A as representing the form of cowling usually fitted to such engines. motor to avoid torsional constraint. Lateral motion of the body was prevented by the V-wires (E) by which it was slung. These were arranged so that, if produced, 202 Chapter 3: The Design Revolution Document 3-10(a-c) 203

they would intersect on the airscrew axis, in order to prevent any lateral forces on As the application of a ring to an engine partakes of the nature of a curve for the the motor from causing the body to roll. Current was led to the motor through turbulence caused by the cylinders it was concluded that any serious difference in mercury cups, M, at the tail of the body and through springs near A (not shown). the degree of turbulence produced by the cylinders on model and full scale would In all tests the normal working range of V/nD was covered. The motor was show itself by a corresponding scale effect on the improvement in drag obtained by controlled on torque, and rotational speed was measured with a chronograph. The adding the ring, which would thus be less effective on full scale than on a model. rotational speed (n) was kept nearly constant over the whole range of V/nD at about A few experiments were therefore made to determine the nature and extent of 43 revs. per second, for most of the tests, and the tunnel speed was varied from 80 the scale effect near the critical value of Reynolds number on a very short cylinder f.p.s. to 35 f.p.s. projecting from a place surface. The results are quoted below. The chronograph was found to be more satisfactory than a stopwatch for mea- Three cases were tested: suring rotational speed. The extra accuracy on n, which affects the results virtually (1)8 in. diameter cylinder 10½ in. long in center of 4 ft. tunnel. in proportion to n3, was considered worth having; with the chronograph the thrust 8 in. diameter cylinder 10½ in. long with false floor nearly in contact with one and torque curves were found to be noticeably smoother than with a stopwatch. end of cylinder. The following tests were made with the modified Bulldog model: (1)Body and engine only with and without airscrew, with various rings. Document 3-10(b), United Aircraft and Transport Corporation Technical (2)Body without engine, with and without airscrew. Advisory Committee, Meeting Minutes, December 1929, Boeing Historical (3)Velocity between cylinders. Rough measurements only for comparison Archives, Seattle, Washington, pp. 213-256. between different rings. PROCEEDINGS OF TUESDAY AFTERNOON, DECEMBER 3, 1929 6.5 Siskin Aeroplane Body with Jaguar Engine (1/5 Scale).—Through the courtesy of Messrs. Sir W. G. Armstrong Whitworth Aircraft, Ltd., who loaned this The meeting convened at 1:30 o’clock .PM. pursuant to noon adjournment. model, it was possible to carry out experiments with rings on an engine having two rows of cylinders, the cylinders in the rear row standing opposite to the gaps CHAIRMAN MEAD: Well, while we are waiting on Colyer and Humphries to between the cylinders in the front row. Viewed from the front this engine presents come in, Mac has given us a sheet of airplane characteristics and I know Chatfield almost a solid disc to the airflow. In spite of this however it was found that a ring and the rest of us have been worrying for some time for a way to get this in a stan- was practically as effective in reducing the drag of this engine as of engines having dardized form that will be acceptable to everyone. It might be a good idea to pass a single row of cylinders. that around to see if we can get constructive comments. Details of the model are shown in Fig. 17. The upper half of the figure shows MR. MONTEITH: Instead of that table of performances, I would rather see the ungeared Jaguar without rocker hats, whilst the lower half shows the geared curves. It really makes two sheets out of it, but you really stop to plot this thing up model, which was fitted with rocker hats. Two typical rings are also shown in the in curves anyhow. figure. As this engine was fitted with an exhaust manifold behind the cylinders, no CHAIRMAN MEAD: That reminds me,—it may work everybody if we have tests were made of this model with rings of the type of Ring A, Fig. 2. these days pretty well filled up, but if you can get your gang together in the matter The following tests were made: of performance testing of course, I have it down for Friday but we may all of talk by (1) Body (with windscreen) and engine only (ungeared), with several rings. Friday we won’t give it the proper consideration,—and see if we can’t light on some (2) Check tests as (1) on the model with geared engine and rocker hats. system that looks reasonable to try out, then we can modify it in six months, but (3) Test with and without spinner and exhaust manifold. here is Northrop testing down at his place and we are testing in Hartford, and we hope Mac is going to be testing sometime, and it will be nice to get down to earth 6.6 Scale Effect on the Drag of Engine Cylinders.—The drag coefficient of an if we could. infinite cylinder is known to undergo a rapid drop in the neighborhood of a certain MR. MONTEITH: Well there is not much trouble in the speeds if we get a value of Reynolds number, Vd/v, lying between the model and full scale values for decent speed course, which we do not have at the present time,—but practically cylinders of the diameter of normal engine cylinders. everybody whom I have run into at all in the industry is making his climbs with It had been suggested by Mr. Relf in Ae. Techl. 399 that though an engine cylin- an altimeter, which is the bunk. I think something we need every place is a good der was very different from an infinite cylinder yet it might be reasonably expected to barograph with a calibration apparatus to go with it if we are going to get any decent exhibit some decrease in drag coefficient near the same value of Reynolds number. climb data at all. 204 Chapter 3: The Design Revolution Document 3-10(a-c) 205

CHAIRMAN MEAD: Last night we had a supplementary meeting over at the speed is not so interesting. Hotel and Mr. Sikorsky was putting up a method of measuring landing speed, and Well, it looks as though we have done about all we could with this matter of that is another thing we are worrying about. transportation. Of course, we would like to analyze this matter of maintenance as MR. MONTEITH: I just had a letter from Reid down at Stanford. He wants us we said this morning, and probably we can in the near future in view of what you to loan him some planes: He has a method of testing landing speeds down there by are doing. photographic methods. He paints one wheel half white and half black and measures the comparisons of performance for this six months with the last six months it with his photographic instruments. we can perhaps make graphically for you when we get the data, and of course you CHAIRMAN MEAD: It would be nice if we could get an instrument [that is] should appreciate that Chatfield and some few people in Hartford are now assigned not too expensive, but also simple and durable we could use on any of our ships to help us on this sort of work so as not to make too much of a burden for the indi- either land or water. Mr. Sikorsky was telling us about a different method that might vidual units. If you can furnish the information, we can sometimes put it in shape be worth while if he would explain that again. and save you some work, but to gain anything out of this data of course it has to be MR. SIKORSKY: Certainly, I would be glad to. You see, here is the size we use in comparative shape all the time and I think we will find that we are going to be (indicating on blackboard). This is a box about actual size. This part here is a piece interested in fewer items than we have discussed this morning that we can substanti- of glass. There is an arm here inside with a visible hand here, and at a place some- ate those figures in the report so we don’t need to take the time to discuss them each where about here,—of course there is an opening in the top to fix this up,—now time, and we can show graphically on a chart prepared ahead some of these things here, there is sticking a support on which a disc of twenty-five centimeters diameter in a very few minutes. is fixed. How, there is existing very reliable data of the pressure on a disc of any Well, the next item I have down for discussion is that of power distribution, believ- dimension may be applied, but it is very easy to graduate it and fairly reliable infor- ing that it was one way to get at the possible improvements in plane performance. mation is available on the pressure at every speed. now, Chatfield had better explain to us his story I think he sent around, and Now the whole thing is to adjust the load here to such an extent as to show your obviously, you can criticize considerably because we have not very much data to desirable speed. Now, of course, all you can get here would be whether your landing work with. is, say better than 54 or whether it is worse. In other words you have to check it a MR. CHATFIELD: I suppose the main objection to this is that based on a small couple of times and sometimes you have to adjust the load a couple of times until machine, a Sperry Messenger which has only three cylinders, and in order to get you get the actual thing, but we make a multitude of tests and I would say that I have it comparable to faster machines, the speed had been stepped up to a speed much the full conviction that the thing is accurate within one mile. For instance, when higher than this airplane actually had. we tested several, or tested the same airplane with several loads then the increase of the results are obtained by a sort of combination or the NACA reports, the landing speed went simply mathematical in accord with what it should do. technical report 304 which gave the flight polars for a complete airplane with four The thing we tested was the ship actually landing. In this case, of course, a second complete sets of wings, and technical report 271 which gave the driver of the com- person must make the observations or it can be tested by the pilot himself. In this ponent exclusive of the wings. The results are nothing more than a combination and case, it is very easy to keep the thing just bouncing back and forth, or even to keep it reconciliation of these two, and I suppose I might put the percents on the board as down. Of course, here it is to keep it down which shows the speed is below a certain being the easiest way of showing them. limit to which the instrument is adjusted. The thing is very simply and cheap to construct. (Mr. Chatfield puts tabulated list on blackboard) McCARTHY: How about the effect of the angle line on the pressure? MR. SIKORSKY: Here is the thing: The slight differences, even a good few this of course, also involves allowance for power lost in propeller so the total degrees,—as far as I remember two or three or five degrees make a very little differ- figure is not a distribution of drag but a distribution of brake horsepower, and it is ence for the disc, therefore, of course, you put the disc, to be normal, to the wind break horsepower accounted for in these various ways. and [approximate] your landing speed. Now there the advantage is, you need no that, I think shows it reasonably well. Of course the propeller is based obviously correction whatsoever, if you do it [at a] thousand foot altitude the thing will show in propeller efficiency and in this case would be 2 percent. you,—the correct speed actually may be 63 miles per hour but it will show you 64, of course one surprising thing which probably is exaggerated as I said, in a very just so the corrections are mathematically taken care of. low portion which is accounted for by induced drag which might be extreme but CHAIRMAN MEAD: The thing seems to me we are interested in landing speed, you might say only 1.2 percent of brake horsepower is going to hold, the rest of it otherwise we do not appreciate the top speed. If the landing speed is high, the top is all absorbed by various resistances and losses, but as mentioned before I put this 206 Chapter 3: The Design Revolution Document 3-10(a-c) 207 account of having jacked up the speed. The induced is normally low and I imagine the picture here and this figure (33.8) certainly is very correct and probably is much would be probably twice at any rate, or perhaps somewhat more than that, but still greater, and yet the bracing and interference that is shown here would permit you to would give the result, the induced drag is a relatively small part of the total. save considerable on some other items. Here the induced drag stays as a low item, of course the most conspicuous item is immediate-interplane bracing and inter- but if you take a heavily loaded airplane,—if you take a big one, if you consider the ference which in a sense is the stuff you can’t find somewhere else. case that you are flying with one unit out of commission, in other words, fairly slow MR. MONTEITH: Pardon me, but in that particular ship do you recall that the speed, in most cases with more multi-engined ships it would be something like 75 or interplane bracing was made of struts? 80 miles per hour, then the figure of induced drag becomes a more considerable one, MR. CHATFIELD: Yes, the interplace bracing was struts in the particular and to take care of it you need span, in fact nothing can be replaced without span. machine, so possibly that again is somewhat higher than normal. In order to give the value of span, I may, for instance, give you the following very I might say the only justification for picking this particular airplane was because accurate information: We built an S-36 amphibian with a gross weight of approxi- it was one from which we could get the data; I don’t think there is any other argu- mately 6,500 pounds, with lots of bad features dynamically, and this ship was pow- ment to be advanced in favor of this set of figures. ered with two Wright motors of 200 horsepower each. With one engine dead it Again, due to the engine this addition is undoubtedly abnormally low because showed a loss of altitude of 400 feet per minute, in other words, an enormous lot of it was a three cylinder engine and we don’t have many three cylinder engines, and altitude. as Mr. Weick has also pointed out, if you went up to nine cylinders not only would this ship had 62 feet span. Now we cut the middle part of the wings, added 5 the total engine drag be multiplied by the ratio of the number of cylinders, but the feet on each side and transposed it into [a] 72 feet span airplane, and with the same drag per cylinder would probably be higher as well on account of the interference. load, in other words, with greater gross weight than we had, we only showed a loss MR. MONTEITH: Well, there is another feature there, Mr. Chatfield, too that of 100 feet per minute instead of 400 feet loss previously. in that particular airplane that was the first air-cooled engine we had you know, and It should be taken for granted that with all other conditions being equal, espe- the idea was to put all the cylinders out in the air. cially in a big ship, you can afford a bigger span if you use outside bracing than if CHATFIELD: It is conventional cowling of the old style with the cowling you will use cantilever sings. This is the thing which to a very large extent permits coming pretty well down to the bottom of the cylinder barrel and not up toward all of us to profit in the externally braced airplanes, and this makes for the fact, as the head as is most common today. Of course, this was all several years ago. a general rule,--I am again speaking mainly of the big ships, the clean cantilever Apparently what goes on inside the engine disc does not make much difference monoplane did not show loss of speed or anything more in general than the braced as regards resistance, isn’t that the general conclusion? external biplane. MR. SHORT: That is what I judge from our tests all right. Again, I say this is true for very huge units. In very small ones, it seems logical CHAIRMAN MEAD: How much improvement do we get? as the little ships must be given very big preferences. MR. McCARTHY: You get one mile more an hour high speed with the mono- now, I hope Mr. Mead will forgive me for the following statement: This is, I plane than you did the biplane. believe that what we actually have to experience in dragging across the air the radial CHAIRMAN MEAD: You must have a lot of miscellaneous in this 33.8, other- engine, of course, the engine does it by itself, we want to get more out of it,—more wise there ought to be a big gain with any monoplane. according to our form of what would be shown by this figure. For example, a very CHAIRMAN MEAD: Have we any data in the family which would be useful or accurate wind tunnel test I made for a very huge engine, the engine nacelle was can we get it anywhere else which would bring this story any more up to date? a little over 5 feet, about 5 ½ feet in diameter. The Jupiter motor was very well MR. CHATFIELD: Mr. McCarthy has one much more advanced than this. streamlined, or rather cowled in so far that the cylinders were sticking [not] much MR. MONTEITH: I might state that in that particular model we made for the here (indicating). As a matter of fact, if you would use the cross section, the engine NACA, I was instrumental in building this set of wings for the test. We had one set would occupy a very inferior part of the circle. arranged with internal bracing so we could put the struts in or take them out, but now in this case, the drag was distributed as follows: This whole business, one- they let Doctor Munk get hold of it finally and he held it so long we never did get third, and the engine cylinder heads, two-thirds. the complete test. We kept trying to find out what that interplane bracing interfer- now, again a very large amount of tests, which were entirely reliable and accu- ence amounted to on the ship but we never got the data. rate because we have several dozens of tests on our airplane cylinders; they show in MR. SIKORSKY: I would like to have the data concerning this picture. The general the following things, that the same cylinder which gives us an L.D. of 11 trouble of course about discussing this thing will be that there would be one equa- will give us the L.D. 18, so here we have a considerable amount of power going in tion with quite a considerable amount of uniform factors,—for example, I think because in this case it is already the difference of the whole airplane plus the induced 208 Chapter 3: The Design Revolution Document 3-10(a-c) 209 drive and everything. Now, however, the windtunnel test showed a considerable MR. McCARTHY: Yes. improvement due to the cowling or a similar combination. For example, the cyl- tail surface 8.7. Landing gear 11.2. Propeller loss 16.1. inder which showed L.D. 11—the best lift drag could be pulled to 14 ½ only by now we designed the same job as a monoplane and at a high speed of 183 miles specification of the NACA cowlings, or rather similar cowlings because we cannot an hour on the same kind of paper and using the same slide rule the wing drag came say on the model what kind of cowlings we put in and how good they would build out 37.8. the actual airplane, so I believe that a very important thing and a very big gain for this first (indicating) was really a sesquiplane, the other a monoplane. us would result from a very careful study of these very cowlings. That is where we Interplane bracing about the same as before. Fuselage 12.2, and radiator 6, tail can save a considerable amount of power. 8.8, landing gear about the same as before, and we have just as good a propeller. Again, I think this is more true for exposed engines, but probably it is largely now here is an O2U-2, a biplane with the famous Wasp engine, and this is cal- correct also for single engine airplanes, and I believe here in this close corporation culated performance and agrees with the actual performance within a mile, 28.9, with the help of the United facilities, we may be enabled to make a considerable step bracing and interference 15.4, fuselage 23.6. ahead because this here is quite a big item in which we can save lots of power. MR. SIKORSKY: These data are at full speed? now, I believe maybe this may be an item which I say is general for all of us, MR. McCARTHY: Yes. If you took it at low speed, your wing drag would go we may expect to use the radial engines on probably most of the designs because up somewhat more. Now it is interesting that the total of this and this (indicating) they are so much more superior and convenient, and the operating cost is less, and agree fairly well with the total drag of the body. The O2U-2 has a large round body people in those cases want them. The [airline] Pan-American, for instance, simply with the gas tank in the system, and these two models (indicating) were made with do not want to talk or to hear about anything except the radial air-cooled engine. small narrow bodies with the gas tank in the body, so the projected area of these jobs That is the situation right now, so it seems a thorough study of all these things may was much less than the O2U-2. be very justifiable and will do us all some good. CHAIRMAN MEAD: That seems to bring again the old problem of air-cooled MR. NORTHROP: One thing might be mentioned in what Mr. Sikorsky had to versus water-cooled drag. I wonder if you can get anywhere with a check on these say relative to interplane bracing, and interference with a large span. I have noticed figures on the basis of what a machine does with retractable landing gear? in the high speed fairly small span planes I have had experience with the take-off MR. MONTEITH: That landing gear on the O2U-2 looks awful small. What and climb are not nearly as good in proportion to the rest of the performance as size wings did you have? high speed, in other words at points where the induced drag would normally we MR. McCARTHY: 30 by 5. The Army test made on a Falcon, wind tunnel test shall say, the ship is highly efficient over a biplane arrangement, but in take-off and on a model and removing the chassis completely they reduced the drag 14 percent, climb it is no better and in some cases, distinctly inferior to biplane arrangement. so that would make the drag on this basis— CHAIRMAN MEAD: Of course, what interests us most is what sort of improve- MR. MONTEITH: Let’s have those figures on the Falcon, put up on the board. ment we could make in those figures. I wish they were more representative of what CHAIRMAN MEAD: Have you anything, Monty, we could put up against this one of our present air engines does, then we would know much better about what as a monoplane with the wheels up? we are working with. MR. MONTEITH: We have the overall test with the model 200. The tests are MR. McCARTHY: Let’s put up some figures which will compare with this. not very accurate with gear in and gear out. We do not figure any performance on CHAIRMAN MEAD: Have you some there, Mac? it though because we have to have the body with just the blunt nose on it. We were MR. McCARTHY: Yes, I have about five or six different airplanes here. supposed to have the NACA cowling you see, so we simply extended the lines back, CHAIRMAN MEAD: Can’t we take an average and set them up on the board? so we have not any idea what the body drag is going to be. MR. McCARTHY: I can put down about three. (Marking figures on the black- MR. McCARTHY: On the Falcon, I have nothing nearly so complete at this. I board). have the drag, the loss to the radiator is 4 ½ percent for the total drag. You would MR. SIKORSKY: Some of these are water-cooled jobs and some are air-cooled. have to divide that by power efficiency, I suppose, that would make 82 percent,— This one is an XO3-1, that is a two-seater observation plane with a geared 1570 that would make the radiator 3.7 percent; the landing gear is 11.5. Those are the motor on it, and estimated high speed of 182 miles an hour. The total wing drag, I only two comparable figures I have. did not separate them, about 35.6. they tried three or four places for the radiator and found the normal position Interplane struts, wires and interference is only 8.3; fuselage is 14. 1. under the body, the old funnel position was the best of all; that was even the best the radiator 6. with this Heindrich type. CHAIRMAN MEAD: Including the engine? CHAIRMAN MEAD: I wonder if it is safe to take those and try and build up 210 Chapter 3: The Design Revolution Document 3-10(a-c) 211 an average set of figures, and then from those figures what improvement anybody MR. SHORT: Now, ours unfortunately thus far have only covered the J-6 300 thought they could get for various reasons? That is, propellers could be improved, (horsepower). I hope before the week is over I will get a report on the Wasp. and engine cowling? Mr. Weick seems to have a lot of dope in his head from his tests We started out the first one following very closely to the rule of the NACA cowl- of what we might be able to do. ing, and the changes which we made however on the J-6 was to put the collector MR. MONTEITH: The thing I am interested in right now is, are we going to ring in the rim of the cowling as the one furnished by the Wright Company is not be able to use NACA cowling on the geared— suitable for strictly NACA cowling. CHAIRMAN MEAD: Well, Ford does. With that first one and with the propeller the Hamilton Company furnished us MR. WEICK: I wouldn’t say he did yet. I spent some time there and made some we did the admirable high speed of 133. tests on that particular ship. He did not have what you would call NACA cowling CHAIRMAN MEAD: Was that in direct comparison with the ship before? on it in the first place, and he didn’t know what it did in the second place. MR. SHORT: I don’t know what the ship would do before, but with some slight CHAIRMAN MEAD: I know the boys finally got it so it would cool passibly, it alternations in the cowling, and with a more suitable propeller on the same ship we does not cool too well; and I was going to suggest it seems to me the McCook Field did 143. type cowling would be ever so much better. CHAIRMAN MEAD: In other words, you lose speed with the coowling? MR. WEICK: Well, I think myself a certain amount of development work has MR. SHORT: With the version of the NACA cowling we had at first. to be done on every installation, especially the hard ones like taking a Hornet for MR. HAMILTON: You improved the cowling? instance, and in the extreme case, taking a geared Hornet on a fairly slow ship and MR. SHORT: We made an improvement on the cowling, namely,—it may be cooling it, and I do not think—that is, I think the general idea is to just say NACA interesting to know how we tried it in here (indicating on the blackboard) cowling is the particular stuff outlined in the report there, but I do not think that is MR. HAMILTON: Is that with the same propeller you got the improvement, what should be stuck to at all necessarily, I think the air should be guided and used or with a different propeller? as efficiently as possible especially in the cases where cooling is hard, and in the case MR. SHORT: With a different propeller. With that token it is hard to differenti- of a geared Hornet for instance I think without any doubt in the World it could be ate even though we do know after the first experiment that the first thought on the made to both cool quite satisfactorily and have a reasonable reduction in drag, but NACA cowling and the exhaust collector ring was not ideal by any means. I do not think you can expect to do it the first time you try; I think you have got to The collector ring came in here (indicating) [as] an attempt to get that effect; that work on that to the point [there] with your reduction in drag; you can use a compara- being the outer foil of the outer wrapper, and this was to develop the supposedly air tively small amount of air but use it in the right place and get effective cooling, too. foil appearance on the wrapper. CHAIRMAN MEAD: I am going on the basis that Tillinghast says the cock On cross section this is a collector ring, and that one,—the collector ring was get- with Breen’s cowling will operate on the ground with the wheels blocked with full ting quite hot, we were not getting the air through there apparently, and with the throttle for indefinite periods—half an hour at a time—without overheating the propeller and that combination of cowling on the nose we got the 133. engine. Well, that is better than we have often heard of with NACA cowling? I don’t We then changed it to this effect (indicating), bridging over here, and leaving a mean that as criticism, but it just happens to work that way that the air seems to be small gap in there and using this whole portion right in here as a cockpit heater; a better used. hot air heater for the carburetor, and with that 1219 propeller design we did on our CHAIRMAN MEAD: He has some control, of course, which the other has not. course, as I said, 142.7. We figured that was near enough to 143 to swing over. MR. MONTEITH: On the other hand; the XP-12 which we built here was just on our third attempt,—the inner wrapper coming off so fashion (indicating), exactly like Breen’s cowling only we did not pull the skirt of the cowling in, we faired we have discontinued it from the rest of the cylinders back to a point ahead of that the job out to give it outside diameter of the engine, and it apparently worked all one (indicating). We had a number of experiments and our Organization went right out here but when he got it back East he had trouble with it overheating. Breen through quite a session on it, and we saw no change of speed by removing that was out here and we laid the the whole thing out with him standing right over us. inner portion, and saved considerable time on construction of course, and then for CHAIRMAN MEAD: I think it all goes to show that we don’t know as much maintenance it was ideal. about this job as we ought to know. now on the first Wasp job we are using some of Captain Breen’s ideas, together MR. MONTEITH: If we use this on the mail line it has to work darn well, not rea- with the NACA cowling, using a fairly well cowled up nose, a single outer wrapper sonably well. I am wondering [if] it is going to be worthwhile to put it on that ship? that is on the Lockheed version like Breen’s, toning it in but using— CHAIRMAN MEAD: I think Short can tell some of his experiences, we saw CHAIRMAN MEAD: (Interrupting) You don’t know how much of this increase that he had been using the tin shears in Wichita. was due to propeller? 212 Chapter 3: The Design Revolution Document 3-10(a-c) 213

MR. SHORT: No, we have not the opportunity now to try that out we have Monty, if we can. changed the cowling. We tried it with this portion out (indicating) and got about MR. MONTEITH: We actually got a gain on P-12. Of course, we had that up 130 miles an hour with that arrangement, just using the collector ring setting out to 183 miles an hour with the engine turning 2170. here in the open and no outer wrapper. MR. SIKORSKY: We had a very reliable information concerning the use of the I think we are safe in saying that on our first ship the NACA cowling did no NACA cowling with a sort of ring around the cylinder heads of our amphibians. good at all. The people who came up from Langley Field to see the job at Washing- In order to have the data as accurate as possible, we made it as follows: We gauged ton said we did well to get any increase, and we heartily agreed with them after we accurately the speed of the ship with the thing on, then we quickly dismounted it played with it ourselves a bit, but it brings out Weick’s statement that everyone is a and then on the same day, the same pilot, the same engines, checked the speed again separate problem itself, and invariably I think it will happen that the first cut at the with the first test, and we found the following data: The ship actually gained more cowling is a disappointment. than five miles per hour at top speed, at least as much in the cruising speed, the dif- MR. McCARTHY: How did the one work out that you had done at Langley ference in the cruising speed appeared to be fairly distinctly visible and very much Field last summer where you had the thing fairing [from] the fuselage? in favor of the NACA cowling, and of course it was measured with the same amount MR. SHORT: The NACA people reported not so well. We have not gotten a real of cruise. final report from them on it. That was faired in around a conventional rectangular now, there was no difference in climb, in other words, I would say that the ship fuselage with no attempt at fairing in the sides, exhausting through gills on the sides. did not lose any in climb, or lost so little that it was impossible to judge it. On our Wasp combination, it points to a very logical arrangement, and as I say, I now, it seems that a little negative difference was in the landing speed which was hope I can tell you what it will do before the week is over. It is a combination of possibly a little bit worse with the cowling than it was without it. Breen’s and NACA, using a nose spinner,—Mr. Mead advised us very timely that we now, this data was rather accurate and certainly was encouraging from the stand- would not get very much engine control over that because the shutter covers only point of using any kind of an arrangement of that sort, because the one which we the very lower portion of the cylinder barrels and probably would not control the used was just simply a first guess which we made, and we tried to be very much on temperature, but it is getting the air in the inlet and that is the problem if we cowl the safe side in order not to overheat the engines, so it was only a tiny ring some- it up too high. thing like fifteen inches wide just covering the tops of the engine cylinders, and We did not want to use Breen’s Venetian shutter effect in there, the maintenance these were the results: The engine did not show at all any signs of overheating, so must be high on that, so we tried to steer clear of that. the cooling was satisfactory, and I believe quite a lot may be gained in this field and MR. NORTHROP: Has anyone any experience in controlling cowling tempera- it seems to be very advantageous to go on further in some manner of search. tures, that is in taking care of too cold weather by decreasing the gap in the rear of As I said, the data of the wind tunnel test we [did] with wind motor amphib- the cowling? ian was still considerably more encouraging because the models with full cowling MR. SHORT: We are trying that now with a slipper ring on the end in fact, it showed improvement of 3 ½ points, in other words 14 ½ total of drag instead of 18. will not only control the temperature but it will control your speed, and we thought CHAIRMAN MEAD: Monty, have you any idea from flying a tri-motored of making a remote control to the cockpit because you will take off, as Colyer plane with the various combinations what the ultimate situation might be? As far as experiences in the mail run, take off from a warm field in Omaha and hit some bad I understood it the outboard motors do not give you any gain, it that correct? weather before you get to Cheyenne, and you have to keep the motor warm, and MR. MONTEITH: That is correct, the center motor gave a gain of 3.1, but the you have to pay the penalty with speed so we had a cam arrangement worked out three sets of NACA cowling added about 120 pounds to the ship, that is almost one on the outer ring so as you would rotate this outer skirt it would creep back and passenger. forth. CHAIRMAN MEAD: I wonder if Weick knows from Langley Field whether MR. NORTHROP: That was a cylindrical skirt in that case? they have arrived at any conclusions about the relation of cowling and motors to the MR. SHORT: Yes. Well, we made all of ours a perfect circle. wings? Isn’t that the reason it don’t work on Monty’s ship? MR. MONTEITH: George, after following up what we said this morning, I MR. WEICK: That is the reason that I ascribe to it, that is you have an inter- would like to point out two things: This discussion of NACA cowling is fine but as ference effect there between nacelles and the wings, and so far in just about every far as any NACA cowling itself is concerned it is not only heavier than the normal case I know of, which is three actual full scale tests, also some wind tunnel tests, the cowling but also increases the maintenance troubles because of the two sets of cowl- interference between a cowled nacelle and a wing nearby has been very great, in fact, ing to take off. great enough to usually nullify any possible gain due to the use of the cowling on the CHAIRMAN MEAD: Of course, the first thing is to get any gain out of it, engines. 214 Chapter 3: The Design Revolution Document 3-10(a-c) 215

the first experience we had with that of NACA was that we cowled the three here, and having large fillets all around, we got an appreciable gain, but by moving engines of a Fokker with whirlwind,—tri-motor whirlwind and we got a gain of this up so that the nacelle came about in this position (indicating) we got less drag about 3½ to 4 miles an hour from the nose cowling, and again got no gain whatever than the sum of the wing drag and the nacelle drag when each was isolated, that from the wing nacelles, and we did not know of course, exactly what to ascribe it to is, the total drag of this was less than the sum of the wing drag of the wing and the but immediately thought of the possible interference because our wind tunnel test nacelle when they were not together, so that it showed that we could get an increase in the twenty foot tunnel had shown just the opposite; it had shown that you get in performance by changing the location of the nacelle and down the projected area more gain from cowling and small nacelle than from cowling and engine in front of the entire works if you had a smooth enough cowling. Of course, if you had an of a large fuselage, and we took two methods then of going on to the problem, one exposed engine in that position you would spoil the lift over a good portion of your was to put strings—small strings a couple of inches long all over the surface on the wing, and the Ford people with their first tri-motors ran into that problem, they had actual Fokker and on all the surface of the nacelle and the bottom of the wing, and their f-5 engines right in front of the wing, and they had such high landing speed we found that just behind the maximum diameter of the nacelle where the nacelle they had to remove them. They took it at the time, it was due to the fact that the started to grow small again there was great turbulence in the first place. The nacelle propeller was in front of the wing. was only about two or three inches from the surface of the wing and the air appar- MR. McCARTHY: (Interrupting) You think if they cowled those in they would ently would not follow through that and down around the nacelle, and most of the be all right? strings in that portion actually pointed forward,—going along over 100 miles an MR. SHORT: Yes, because we did not get any increase in lift by doing that, hour. It seemed very queer but it showed the condition of the flow. none whatever. We only did it at low angles of [attack]. I do not know what would then, in the wind tunnel tests we took, in order to get as near as possible to the happen at high angles of [attack]. full schedule figures, we took a section of the Fokker wing and we built up a nacelle MR. HAMILTON: High angles would be interesting as far as your landing just like the Fokker nacelle as near as we could do it, and with a small imitation of speed is concerned, and that is where Ford had trouble. a motor. We got about the same drag coefficient for that nacelle as we got for the MR. WEICK: There are no tests giving reliable data in this Country. There have complete nacelle, so we knew that part was just about right, and we found with the been a few small tests made in Germany, but they also did not give just exactly what cowled nacelle we had just as much drag when it was next to the wing as we did with you want. Right at the present time,—in fact the last thing I did when I left NACA the uncowled nacelle, and then we did two things; first we started fairing that into was to outline a series of tests they are now working on and not in the tunnel yet, the wing to see whether if by proper fairing we could reduce that interference, and which consists of tests on a combination of a wing and a propeller in a nacelle, we found in that particular case we could reduce it very materially, and by putting and the nacelle has forty different positions. Approximately forty different positions on a large fairing we increased the actual front projected area and appreciably we with respect to the wing using the propeller as a pusher that includes tests with the could reduce the drag very much so that we got almost the entire gain that we had propeller entirely above the wing and entirely below the wing and three different anticipated we might get neglecting interference entirely. positions before and aft, so that it is right as near the wing as it could get clearance, CHAIRMAN MEAD: That is simply from the cowling into the wing? and then some positions fore and aft, and in some cases like that where it is built MR. SHORT: Simply fairing from the nacelle which was already covered with right into the wing and other cases where the nacelle is suspended above or below NACA cowling clear to the wing without changing the location of it, and inciden- the wings. tally at that that type of fairing was put on the ship and a total of increase of twelve CHAIRMAN MEAD: Monty, are you going to have NACA cowling on the 200? miles an hour was finally obtained. Also, the nacelle was moved in the model but MR. MONTEITH: That is what we had planned originally, yes. it was too much of a job to do that full scale, but on the model, the nacelle was CHAIRMAN MEAD: Based on what we know now, the trouble with NACA moved to several positions above and below the wing and the only one which was cowling is heating, isn’t it? moved was the nacelle with the complete NACA cowling, and it was found that by MR. MONTEITH: I am still worried about that geared engine though. moving it far enough down you could get away from fairing it, but it had to be an CHAIRMAN MEAD: You mean in the 200 you are going to use geared engine? appreciable distance below the wing before the interference would be noticeable. MR. WEICK: With the cowled nacelle, which is the only thing we had there as I the best results were obtained by putting the nacelle exactly in front of the wing remember it, if you got down close to one engine diameter you could get away from (indicating on the blackboard) with the top of the nacelle parallel to the top of the the interference, but that is more than you can ordinarily get in an airplane. wing. [It] will just take a small space here (illustrating). Supposing that is the MR. MONTEITH: You mean the space to be one inch in diameter? wing (indicating), and the normal nacelle came in about like this, and when it was MR. WEICK: Supposing the diameter of that engine would be say 45 inches in that position we got no gain whatsoever, because the air would not flow through below the wing, that is what I mean. 216 Chapter 3: The Design Revolution Document 3-10(a-c) 217

MR. MONTEITH: On that question of Mr. Egtvedt’s here a moment ago, would not let us have anything further if they discovered that we were broadcasting Doctor Raam made some tests four years ago on propellers in front of large wings. it. It would get them in bad. The wings were much larger than the ones we are using however. CHAIRMAN MEAD: You would not get any more? MR. WEICK: Yes. Well, his tests, I think, were very good and they show one MR. WEICK: Yes, that it is, sure. thing too, they show in a case like that, or in any case where the propeller is in front CHAIRMAN MEAD: Then, as far as I can see at the moment we can’t help of the wing, you have to put that appreciably in front of the wing or you get a very Monty very much except as to additional sealing in the engine. I would like to know noticeable loss, and from his tests I got an idea on an ordinary arrangement, we will in what regard whether the cylinders themselves are warm, or whether you are just say a nine-foot propeller, which was what we had on the Fokker that we were work- guessing by the oil temperature? ing with at the time, if we had put the nacelles up into the wing such as that and MR. MONTEITH: The cylinders themselves are warm on the transport, up as then put the engines far enough ahead so that the propeller was approximately 18 high as 580. inches ahead of the leading edge of the wing instead of just I think about five or six CHAIRMAN MEAD: How about the barrels? inches as they were in the actual installation, there would have been an appreciable MR. MONTEITH: The barrels are running in proportion; down around 470. gain all the way through, and it looks to me off-hand as if that is the best condition CHAIRMAN MEAD: I should think Breen’s scheme of blocking in between the considering high speed. Now, what that particular condition would do to the lift at cylinders may improve that somewhat? high angles or landing speed I do not know. MR. WEICK: I think there is no doubt in the World that some scheme of block- MR. HAMILTON: There is a Russian reproduction of the Ford tri-motor which ing in between the cylinders and putting the air that you take in right where you I saw in London, where they have the engines out from the wing about better than want it instead of testing it simply filter through in the easiest possible manner, if it two feet and it is not cowled. It has a Lynx engine and they claim they have tried the is guided to where it should be that would undoubtedly cool better than even let- position of the Ford and also this position and the landing speed is no greater in this ting the engine stick out in the open, but that does add such things as maintenance position than with the Ford,—than with the low wing position. However, their top troubles which of course are already objectionable. speed is considerably improved. It seems to bear out the information due to the fact CHAIRMAN MEAD: Do you think, Monty, we could do you any good in they are I should say almost 30 inches ahead of the leading edge. Hartford with the 40-B? Of course we are plumbers when it comes to cowling, but CHAIRMAN MEAD: Well, it strikes me one thing we would like to know as we can patch up something,—it has a geared engine in it,—to try and fix up some soon as possible is the results of these tests in NACA laboratories, with the propeller installation which will cool there,—help anticipating trouble with the 200, or do turnings. you think the ships are so different that we would not get you very much of any MR. WEICK: It is true, but you won’t get those for some time. valuable information? CHAIRMAN MEAD: I wondered if there is any chance of you Mr. Chatfield; MR. MONTEITH: Well, the ship is a little bit slower than we suspect the 200 somebody who knows them well, to get in there ahead of the rest of the gang? to be so that will be an additional trouble to overcome, but what I am wondering is MR. WEICK: Well, there is that possibility— whether or not the additional speed we hope to get out of the cowling will overbal- MR. HAMILTON: (Interrupting) They tell me, just with that idea in mind— ance the additional weight and additional cost? MR. WEICK: (Resuming) Actually, I just did go down and got all the dope they CHAIRMAN MEAD: I wonder what Weick thinks about that? That is a pretty have done on high tip speed so far. Those tests were started by me and they just big body. Do you have any information which would check that? finished them up after I left and I had no difficulty whatever in getting that stuff MR. MONTEITH: The body is the same diameter as the engine. because I know the fellows down there and because also I was in on the start of the CHAIRMAN MEAD: Well, it is bigger than anything we have had as a single thing and they realized that I had something to do with them down there. engine ship. now it so happened that I was in on the start of this particular thing and will MR. MONTEITH: Well, the over all diameter of the J-5 is approximately the have no difficulty whatever getting that information by going there, but I can’t get same diameter as the engine, it does not look like it, but it is. it by getting them to write it to me. MR. WEICK: The over all diameter of the open-cock fuselage in the 20-foot wind CHAIRMAN MEAD: Then if we carry out our present scheme of getting such tunnel test was also approximately the same as the J-5 engine used in those tests. information to Chatfield he can broadcast it through the news letter and everyone MR. McCARTHY: On the F2U, Monty, the body was almost exactly the diam- will get it about, as quickly as possible. eter of the engine with the NACA cowling they put on it, the top came right directly MR. WEICK: Only, any such information which comes out before it is pub- back and the sides were probed in only a very small amount. We copied Breen’s stuff lished by NACA should be kept within the family because otherwise they, of course, almost exactly; had these deflectors behind the cylinders, and we were right at the 218 Chapter 3: The Design Revolution Document 3-10(a-c) 219 danger point as far as cylinder temperature is concerned. have if you said from ten to fifteen, this rather proves it would be better to be down MR. MONTEITH: On the 80-A we tried the NACA deflectors on the cylinders here at ten, wouldn’t it? but it made no difference. MR. HAMILTON: You are dealing now with only top speed, how about CHAIRMAN MEAD: Mac, have you had one of your jobs with NACA on it? climb—rate of climb? MR. McCARTHY: Yes, the XA2-1. MR. MONTEITH: We did not have it on the XP12-A so we don’t know. CHAIRMAN MEAD: I don’t see quite why Monty feels he is not going to get MR. WEICK: What you expect, of course, is if you get proper cooling and gain in speed based on these other tests. proper power out of your engine, and if you get the decrease in drag which you MR. MONTEITH: I don’t say we won’t gain it or won’t get the gain. We have expect from your cowling, you would expect it to take slightly less power to drive managed to put about 11 miles on the mail plane by both fairing out the body it through the air at climbing speed and therefore get a little better climb; but it behind the engine and putting the NACA cowling on, and it did run hotter, which certainly would be a small amount. from Mr. Colyer’s point of view, is not so good. We got an appreciable increase on CHAIRMAN MEAD: I would like to swing this around on one other track, and the XP12-A, and on the Transports, we failed to, except the nosed engines. that is do we know what that fuselage proof would be with a wing radiator? MR. MONTEITH: I would like to go ahead with the 200 with the NACA MR. MONTEITH: That would depend on what kind of ring radiator you used. cowling on because we have started out that way, and if we get into trouble we can CHAIRMAN MEAD: Air-cooled engines do not look so bad on this method of change to something else. scoring, but the question is how accurate it is. CHAIRMAN MEAD: We seem to be a long ways from this story, but what I MR. WEICK: The three fastest ships at Cleveland apparently had radial air- was driving at here is fuselage complete with engine runs 20.18 and 23 which shows cooled engines. apparently that air-cooled and water-cooled do not vary a whole lot. Here is an air- CHAIRMAN MEAD: Of course, we get to this top but it certainly looks from cooled at 23, and water-cooled at 20, and if we should take 20 as an average, which Chatfield’s figures, which I think Monty has probably checked up by now to a cer- might be safe, how much reduction would we get on cowling in its best possible tain extent, that the little engine which can be completely cowled is going to give form? Say I can put 20 in this column of averages, what would we have over there the big engine something to think about. (indicating)? MR. SIKORSKY: Yes, I believe so, because you see actually the engine interfer- MR. WEICK: According to our tests you get a noticeable decrease. ence enters not to so say one time, but enters in the cube at least because the engine MR. McCARTHY: It shows six or seven percent decrease in the fuselage drag,— interference and size, which in our case, is to put the wing much higher than we the total drag. want and therefore have longer struts and so on, and therefore this thing may be of MR. WEICK: The total drag, which would be, say somewhere between ten and very considerable value because the actual gain would be much more. fifteen for that drag. CHAIRMAN MEAD: On this item of landing gear, is it going to be correct to MR. MONTEITH: Here are three figures which may help you, George: assume that is entirely washed out if we can have a retractable gear, or is it going to the equivalent flat plat area of model 83 which was the forerunner of the B-12 appear somewhere else in this picture? figured out to be 6.58 square feet. The B-12 figures 8.34, and XP12-A figures 5.89. MR. MONTEITH: It depends entirely on how you retract it. On the 200, we CHAIRMAN MEAD: Just a moment, will you give me those figures again, don’t retract it at all, as you saw from the lockup last night. Monty? MR. WEICK: Then, there is another thing, a good many ships you would have MR. MONTEITH: The 83, 6.58; the P12, 8.34; the XP12-A, 5.89. That is to build some special,—you would have to make some special changes to get room calculated from flight tests from the high speed. Take the wing characteristics which to put the landing gear when you retract it. were alike in each case and simply figure out your wing horsepower and parasite CHAIRMAN MEAD: That is what I say, if we put it down completely at zero it horsepower and back. is going to be there in form. now the difference between the 83 and the P12, of course, lies in the wheels, MR. MONTEITH: If you build a ship with retractable landing gear on it you landing gear and streamlines behind the cowling, and little Military stuff like sights have to start with the gear and build the ship around it. and things of that kind. CHAIRMAN MEAD: From the looks of those figures, it looks as though 11 the difference between the P12 and XP12-A is due to the NACA cowling alone. would be a fairly decent average for landing gear, and I was wondering if we dared to There is some question about that last figure because I had to estimate what the put down something like 4 or 5 on retractable landing gear on a basis that although power of that engine was at 2,170 r.p.m. that item might be wiped out it might be somewhere else equivalent to that? CHAIRMAN MEAD: It would be better to run this down then more than we MR. MONTEITH: You can wipe out about eight out of eleven,—make it three. 220 Chapter 3: The Design Revolution Document 3-10(a-c) 221

CHAIRMAN MEAD: The next question is,—of course, the big item seems to MR. WEICK: It is going to be almost exactly the same unless there is tip speed be this old offender, the propeller. loss to start with. MR. SIKORSKY: This item is a very interesting item, and I would like to find MR. SIKORSKY: What is the practical efficiency for, say around a cruising speed out actually how many give us the proper full amount of revolutions of 80 percent of 110 or 115 miles per hour and a top speed of 135 miles an hour? for, if so, it is too bad because there is no room for improvement. MR. WEICK: Well, I will take from those wind tunnel tests,—I will make a MR. MONTEITH: If everything was as efficient as the propeller, we would be guess at somewhere around 77 or 78 percent. in fine shape. I think we better leave the propeller manufacturers alone for a while. MR. SIKORSKY: 77 or 78? As high as that? MR. WEICK: According to our wind propeller test you can run a propeller up MR. WEICK: Yes. That is cruising at 110 which is just about exactly the same to 950 or 1050 for the tip speed without any depreciable loss, and from then on, the efficiency you would get at full speed if your tip speed was not above a thousand, or loss is very peculiar, it is almost lineal. to be perfectly safe, 950; that is perfectly safe. on ordinary installations with direct drive air-cooled engines, you can get effi- CHAIRMAN MEAD: Well, Northrop has another scheme, or course we would ciencies, that is, propulsive efficiencies, not allowing for propulsion efficiencies, like to hear about from the propeller folks, whether you really think this big gain in including the effect of the propeller on the body; you can get efficiencies as high as putting the propeller behind the wing as he has,—he has buried the engine and has 84 or 85 percent with the highest pitch propeller, and 80 to 85 for any propeller nothing in front of it as he gets the wing thicker he hopes— which does not run at too high tip speed and which is on a ship going say 16 miles MR. NORTHROP: (Interrupting) Our scheme was utilized really to get more an hour faster. efficiency out of the wing rather than the propeller and we plan to try it both ways CHAIRMAN MEAD: (Interrupting) If you give the propeller the best break for that reason. We did not know for sure which way would be the best. We felt the and put it in the proper relation to the wing (I don’t know what that is myself) and speed of the propeller was such it would be less effective by comparatively smooth run it at the correct speed what is the very best efficiency anybody dares to figure for flow of the wing than the wing itself would be affected by the turbulence caused by propeller? the propeller so close in front. MR. HAMILTON: Eighty percent. CHAIRMAN MEAD: Of course, these figures we are putting down here are going MR. WEICK: It depends on your figuring, of course— to affect something until we get 100 percent again. Incidentally, did they tell you that MR. CAIDWELL: Well, we have made tests of ninety percent. the Travel Air folks told Short they gained 60 miles an hour with the cowlings? MR. WEICK: The very best we have ever gotten in the full-scale funnel, that is MR. MONTEITH: I don’t believe it. with a geared propeller which ran at quite low tip speed and which was large with MR. SHORT: Do you expect any propeller design change due to the cowling on respect to the body, a geared propeller on a J-5 engine; a very high pitched geared the front? propeller, we actually got one case of eighty-nine percent. MR. MONTEITH: No, I do not expect that would make any change [to the] CHAIRMAN MEAD: I would hate like the deuce to give you 525 horsepower propeller. and have you throw away a hundred just like it did not cost anything. MR. SHORT: Don’t we in effect have a wall there equal to the outside band to MR. WEICK: The thing is if you don’t want that done, if you don’t want us to the box the propeller is working in? throw away fifty, then you have to design the engine and airplane to get the most MR. WEICK: Yes, but actually it is in a portion which is very close to the hub efficient propeller. as far as the whole propeller is concerned and which is comparatively unimportant CHAIRMAN MEAD: That is what I am talking about. This is our “hope to as far as the whole propeller is concerned, and the average propeller of the present be” column over here (indicating), and that is what I was wondering, whether you day is designed so that the pitch should reduce in towards the hub anyway because felt that you could do materially better than this; whether it is safe to put down any of the short air velocity with any body; that is, an ordinary air-cooled engine with- such figures, say, as ninety percent, which actually is correct, I think for certain high out cowling has a great effect in reducing the velocity of the air through the central speed motor boats working in the water. portion of the propeller, and so had the NACA cowling type, and there is not a MR. WEICK: Of course, there is one thing, you are talking here about the high very great difference there between one and the other as far as the propeller is con- speed condition of the airplane which is not necessarily the most important condi- cerned. tion at all. The propeller efficiency at take-off is very much lower and the propeller Incidentally, we have made surveys in the plane of the propeller with both types efficiency at climb is also much lower than the propeller efficiency at high speed. and the air is actually slowed up more in a sort of uniform manner with the NACA CHAIRMAN MEAD: Aren’t we interested from Colyer’s standpoint of the pro- cowling than it is without. Without it, it is slowed up as much but in a different peller efficiency at part throttle? manner. It does not go out to that ring, it goes off in between, but I do not think 222 Chapter 3: The Design Revolution Document 3-10(a-c) 223 that just that would bare any appreciable effect on propellers any way. The only there, and you get a very noticeable increase; it starts at tip speeds of around 1000 effect it could have that I could see would be on the pitch distribution, that is the on the average, and of now as, in one case 950, and most of them are between 1000 distribution of blade angles and that is not critical in the first place, and it is not and 1050, then it goes down fairly rapidly. greatly different than what it would be without the cowling in the second place. CHAIRMAN MEAD: I was getting very discouraged over this gear situation. CHAIRMAN MEAD: Should we have zero for interplane bracing and interfer- We finally got some gears which would stand up and you gentlemen over here did ence if we had a monoplane without supports? not seem to get any gain except in the seaplane, that is, the high speed situation did MR. MONTEITH: No. not improve very much and so on. Lately though, I seem to be considerably cheered MR. WEICK: The trouble is you would have more profile drag. up by this noise situation which will probably require propellers whether we get any MR. MONTEITH: You get a certain amount of interference between the body gain or not. and the wing. It is a minimum in that case, but it is not wiped out. I do not know how others feel about that, but riding up on the ship the other CHAIRMAN MEAD: Certainly these figures are much better than those which day, which as I said before was the best transport we ever have been in, its principle we had to begin with when we started our investigation, and Mr. Chatfield is going deficiency is noise, and certainly we can quiet the exhaust down some but it seems as to keep right at that because as fast as we get information we would like to add it to though a good deal of the noise over turning fifteen or sixteen hundred was from the this information and make it as useful as possible. propeller. I don’t know how you feel about it but it seemed to me that one situation MR. SIKORSKY: I would like to ask the following question; In a single or multi- was the slowing down of the propeller. engine ship around 500 horsepower, the motor and installation in weather will give MR. EGTVEDT: I would like to ask one question too: What would be the effect you approximately 200 to approximately 250 pounds more weight per each point of this problem of Mr. Sikorsky’s here with the variable pitch propeller? Now right unit. on that particular problem of gearing as against direct drive with a propeller of that now in the multi-engine, or say any heavy Transport ship we are concerned with kind— the drag from the very beginning, for instance, in the flying boat we are concerned MR. CALDWELL: (Interrupting) Well, I think the variable pitch will give a with the drag of the propeller at 25 miles per hour, then with the drag of the propel- decided improvement all right on the static thrust and the efficiency, or rather the ler to the pull all the way further to the take off speed, then to a certain slow speed power at the take-off. I think it would be intermediate though between the gear and like say 75 or 80 miles per hour when we carry very much and about the full effi- the direct drive—the fixed pitch would probably be lighter; a great deal lighter. ciency of the propeller as high as possible in order to keep a good flight, and at the MR. EGTVEDT: It might be possible, for instance with a land plane where the present time the operating company always requires us to give a high ceiling with conditions are not quite as severe as has been pointed out for the sea plane, or flying one power unit out of commission and therefore, we have another point where the boat, that that would more than off-set our increased weights and other difficulties. propeller efficiency is a very high volume. Then finally comes the reducing speed, On the other hand with a flying boat it is very doubtful whether it would take care then finally, most important of all comes the top speed. of the difference. now taking as a reasonable average of all these conditions, what do you gentle- CHAIRMAN MEAD: I think we ought to consider this propeller in the most men think would be our best guess for the future? I am speaking now of, say one advantageous position as well as the right speed, and we have only tried one experi- or two years from now, should we ask Mr. Mead for a direct drive engine of 500 ment in that regard which probably was not conducted very accurately, but there horsepower or 575 horsepower, or should we ask for a geared engine, and if geared, did not seem to be much gain in putting the propeller six inches further ahead, what ratio? Let’s see what would be the answer to that? and possibly we should have gone further yet but I think it is worth several pounds MR. CALDWELL: I do not think there would be any difficulty to us. Probably in the engine if the propeller efficiency can be stepped up somewhere. We found that sort of step would be a very big increase in efficiency. At the top speed I believe according to those tests it is one mile an hour, and it cost us nearly twenty pounds we figured direct drive Hornet would be about 1100 feet a second which would be in the engine to get it and that did not seem to be very worth while. beyond the point of good condition for tip speed. MR. HAMILTON: Mr. Mead, it seems to me there is so much involved in of course, with a metal propeller we are able to go a little further in tip speed on this matter of proportion and its effect on future design that I would like to get account of thin sections that it is possible with the wood propeller. these propeller experts organized tonight to give you some idea of what we have in The tunnel tests with rather thick sections of used propellers have an extreme loss mind for next year, tomorrow, and then you ask us a lot of questions and we will of efficiency from 18 percent down to about 42 percent, going up to 1500 feet a try to give you some where near an answer or they will, rather, not me, but the second,—but with metal propellers, we can’t go very much higher. thing always comes back to the propeller I find after the plane is all finished and MR. WEICK: NACA have just completed several tests they started [when I] was the design is completed you take it out and expect the propeller men to put on the 224 Chapter 3: The Design Revolution Document 3-10(a-c) 225 miles per hour that were laid down on the paper, and I feel that we might be able to change it around, but it is more of a problem than any of us thought; you just can’t contribute, but when they talk about ninety percent efficiency and all that sort of take some rule of thumb and do it, which seems to indicate to me that we just don’t thing I just wonder if these fellows realize that later you are going to be asking for know as much as we ought to about the job. some of those things, therefore I would like to give you an outline tomorrow of what MR. HAMILTON: I am fairly convinced of that. we have in mind. MR. WEICK: Incidentally, in connection with that, I got a chance to talk with I was not here yesterday, but we would like to ask some questions about what you Mr. North quite a while one evening—he is the party in England who has been want and why you want it, and then we will try to give you some idea as to what can doing most of the work along with [Townend] down there, and he said very decid- be done with the propeller proposition as far as gearing, lighter weight and all that edly—and I was glad to hear him say it—that you could not expect to get any gain sort of thing. out of a Townend ring unless you did a lot of preliminary work. That is, the way he I think unless you go at this propeller proposition as one problem, itself, and just worked it was to put a test—put a model in a wind tunnel and then start putting have it tied up to this matter of the over all efficiency of the plane, why, we are not Townend rings on it and small scale stuff. It was not what might be the best actually going to get much out of it from a propeller point of view at least, I have not gath- in full scale, but it was the best he could do, and he figured that if he could get it in ered much from listening to all this so far; we are just sparring back and forth and I about five or six tries of different Townend rings and different modifications then he do not see that we are getting much information out of it. We are asking questions could put the results of his work into a full scale job and hope to get some increase, now that cannot be answered off-hand intelligently. but just taking a clear shot out of the sky you could not tell what you could expect CHAIRMAN MEAD: We are just figuring an average of figures, and you all to get, and the same is true of the NACA cowling. seem to agree that between 80 and 85 percent is where you will land, and that is a good place to begin because we have talked sometimes and thought “if we only had 18 percent”. Well, you seem to all agree you will do better than that, and what Document 3-10(c), “The Curtiss Anti-Drag Ring,” Curtiss-Wright Review 1 we intended to do tomorrow was talk transport ships and Military ships and con- (December 1930) sequent improvements in performance, and from that, try to lay down the program for United as far as research went on all these various items, and I was simply trying CURTISS-WRIGHT REVIEW to get down here some average which looked reasonable and everyone agreed to, December, 1930 and then of course looking off into the future if anybody was willing to commit themselves as to what possible improvement they might expect. The Curtiss Anti Drag Ring You see, we have been wondering if you really felt you could get a gain of ten percent, that is, run up from 80 to 90 percent efficiency, why, that was going to help Greater efficiency—more speed with the same horsepower—is a problem on a lot if you honestly feel, as you apparently do, that we can only take up 5 percent, that is something to figure on, and to have to look somewhere else to get a big which the airplane industry is concentrating today. gain. The most effective way to achieve greater efficiency is by the reduction of para- MR. HAMILTON: This matter of propulsion is so closely tied up to what site, or unnecessary air resistance. is immediately behind the propeller, and according to Mr. Weick’s experience in Other things being equal, the air cooled radial type of engine, with its short NACA, and also on account of the fact that people cut out a dishpan and put it in crankshaft and small crankcase, is lighter than any other type. But when the parts front of their engine and think they have an NACA cowling, we have some ideas as of an airplane which offer an undue resistance to passage through the air are under far as United is concerned and as far as the industry is concerned. consideration, the popular radial engine must be considered one of the worst offend- now, we have some ideas here, and we wonder whether we can utilize Mr. Weick’s ers. Hence, a great deal of attention is being given to the problem of cowling this experience to the best advantage by undertaking some of this cowling work? type of engine, so as to make it more efficient from the standpoint of drag. As a CHAIRMAN MEAD: I think you can’t interest the rest of us any more than by practical solution of this problem, the Curtiss Anti Drag Ring has been developed taking that job over. in the Garden City Laboratories of the Curtiss Aeroplane & Motor Company. This MR. HAMILTON: We are not the least bit unmindful of what is involved in device will be manufactured in the Garden City Factory and will soon be offered to such a nasty mess as to try to assist in cowling. the industry for use on all makes of radial engines. CHAIRMAN MEAD: What I gave up on, when you hear from around the table The appearance of the Curtiss Anti Drag Ring is best made clear by the accom- none of these ground rules apparently laid down work in each case and you have to panying photographs. It is a short barrel-shaped form of sheet duralumin surround- 226 Chapter 3: The Design Revolution

ing the engine cylinders. It has the same contour its the upper surface of the Curtiss C-72 airfoil (wing section). Experiments first made in England by Mr. Townend had shown the possibility of reducing drag with a ring of this type. The problem then, was to investigate the theory of this type of cowling—how and why it works—and thereafter to find out how to design and install the ring for the best results, both aerodynamically and from the standpoint of maintenance. This development from the Townend ring has resulted in the Curtiss Anti Drag Ring. To obtain the necessary information, a one-quarter scale wood model of the fuselage of a new Curtiss airplane was built. Models of Wright radial engine cylinders were attached to the fuselage in their proper positions. A number of rings of different shapes were constructed, some spun out of brass and some built up from wood segments. The fuselage model was then set up in the Curtiss seven-foot wind tunnel, and drag tests were made. The wind resistance was measured without any ring and with each ring in place, trying different fore-and-aft locations for each one. In this way the best shape and location for the ring was determined. A large number of other tests were made to find the effects of various changes and to clear up questions of theory. Air is violently deflected outward away from the fuselage when it strikes the projecting cylinders of a radial engine. This breaks down the smooth flow of air over the entire fuselage and leaves a turbulent wake much larger in diameter than the engine The anti drag ring prevents the turbulence caused by the cylinders from spreading out, and forces the air behind the cylinders in toward the fuselage, iron- ing out irregularities in the flow. The Curtiss wind tunnel tests showed that a drag reduction of forty-six percent was realized from these effects. It is an interesting fact that there is actually a forward-acting, or anti-drag, force acting on the ring, a force which may amount to as much as two hundred pounds on a fast airplane. This force is due to the angle at which the ring is set with relation to the direction of airflow over the nose of the fuselage. As finally developed, the Curtiss Anti Drag Ring is simple and light in construction, does not require any special supports, and is easily removed for maintenance work on the engine. It has been used with great success on several Curtiss Airplanes. Speed increases of nineteen miles per hour have been obtained, and it has been found that the cooling of the engine is improved. Document 3-11(a-b) 227

Document 3-11(a-b)

(a) United Aircraft and Transport Corporation Technical Advisory Committee Minutes held at the Pratt & Whitney plant, Hartford, CT, 19-23 May 1930, pp. 230-238, in Archive and Historical Resource Center, United Technologies Corporation, East Hartford, CT. [Courtesy F. Robert van der Linden.]

(b) “The Reminiscences of Harold Hicks,” August 1951, transcript, pp. 61-69. Henry Ford Museum and Greenfield Village Research Center, Dearborn, Mich.

This pair of documents reflects the U.S. aircraft industry’s growing interest for data about the aerodynamic value of engine nacelle placement around 1930. The first comprises an excerpt from the May 1930 meeting of the United Aircraft and Transport Corporation Technical Advisory Committee, held at the Pratt & Whit- ney plant in Hartford, Connecticut. At the time, no aeronautical organization was larger than the United Aircraft and Transport Organization. Just newly formed in 1930, the organization consisted of four major airplane manufacturers: the Boeing Aircraft Company of Seattle, Washington, which recently had acquired the Stea- rman Company of Wichita, Kansas, and the Hamilton Metalplane Company in Milwaukee; the Vought Aircraft Company of Long Island, New York; the Sikorsky Aircraft Company of Bridgeport, Connecticut; and the Northrop Aircraft Company of Burbank, California. It also consisted of a leading engine manufacturer, Pratt & Whitney, which had just come out with two excellent new radial air-cooled engines (the Wasp and the Hornet); the largest American propeller manufacturer, Hamilton Standard, which resulted from the merger of Hamilton’s propeller company with the Standard Steel Propeller Company of Pittsburgh, Pennsylvania; two airlines, United, which ran from New York through Chicago to San Francisco, following the old air mail routes, and Pacific, which ran from San Francisco up to Seattle. Two other combinations of companies had been formed at about the same time, but United was the largest and most profitable. The excerpt reproduced below from the minutes of the May 1930 meeting of United’s technical advisory committee is extremely interesting from many vantage points. It captures an exchange between United executive George J. Mead; Boeing’s Charles N. Monteith, formerly an officer in the engineering division of the air corps at McCook Field in Dayton (he had written a book in 1924 called Simple Aerody- 228 Chapter 3: The Design Revolution Document 3-11(a-b) 229 namics and the Airplane); and Fred Weick, who had just moved from the NACA to The first experience we had at the NACA was that we cowled the three engines Hamilton. The conversation definitely highlighted that the industry did not fully of a Fokker tri-motor with Whirlwind engines and we got a gain of about 3 ½ to recognize the importance of NACA’s research to transport design—something that 4 miles an hour from the nose cowling, and again got no gain whatever from the would quickly change over the course of the early 1930s. In fact, operating prob- wing nacelles, and we did not know, of course, exactly what to ascribe it to but lems caused by placing engine nacelles under the wing of a multiengine transport, immediately in the twenty foot tunnel had shown just the opposite; it had shown the Ford trimotor, were specifically mentioned during this part of the meeting. that you get more gain from cowling an engine in a small nacelle than from cowling The second document, from the reminiscences of engineer Harold Hicks, the an engine in front of a large fuselage, and we took two methods then of going on to chief designer of the Ford Trimotor airplane, also testifies to the industry’s growing the problem, one was to put strings—small strings a couple of inches long all over concern around 1930 not just for help with proper engine nacelle positioning, but the surface on the actual Fokker and on all the surface of the nacelle and the bottom with proper aerodynamic design data overall. The design of the Trimotor has always of the wing, and we found that just behind the maximum diameter of the nacelle been portrayed as an empirical and unsophisticated process and a stunning example where the nacelle started to grow small again there was great turbulence in the first of the old methods of designing aircraft during the 1920s. Hicks’s recollections illus- place. The nacelle was only about two or three inches from the surface of the wing trate that Ford aeronautical engineers were aware of the disadvantages of placing the and the air apparently would not follow through that and down around the nacelle, nacelle below the wing, and they were considering the use of NACA developments and most of the strings in that portion actually pointed forward; in other words here to alleviate the aerodynamic disadvantages. The fact that they were willing to incor- we are going along with strings pointing forward, going along over 100 miles an porate new innovations into a less-than-state-of-the-art design highlights the inde- hour. It seemed very queer but it showed the condition of the flow. terminacy in design during the period. The interview with Hicks was conducted Then, in the wind tunnel tests we took, in order to get as near as possible to the in August 1951 by Owen Bombard of the Oral History Section of the Ford Motor full schedule figures, we took a section of the Fokker wing and we built up a nacelle Company Archives in Dearborn, Michigan, and brings the reader a valuable insight just like the Fokker nacelle as near as we could do it, and with a small imitation of into the design logic behind one of the most well-known aircraft of the period. a motor. We got about the same drag coefficient for that nacelle as we got for the complete nacelle, so we knew that part was just about right, and we found with the cowled nacelle we had just as much drag when it was next to the wing as we did with Document 3-11(a), United Aircraft and Transport Corporation Technical the uncowled nacelle, and then we did two things; first we started fairing that into Advisory Committee Minutes held at the Pratt & Whitney plant, Hartford, CT, the wing to see whether if by proper fairing we could reduce that interference, and 19-23 May 1930. we found in that particular case we could reduce it very materially, and by putting on a large fairing we increased the actual front projected area appreciably and we CHAIRMAN MEAD: Monty, have you any idea from flying a tri-motored could reduce the drag very much so that we got almost the entire gain that we had plane with the various combinations what the ultimate situation might be? As far anticipated we might get neglecting interference entirely. as I understood it [N.A.C.A cowling] the outboard motors does not give any gain, CHAIRMAN MEAD: That is simply from the cowling into the wing? is that correct? MR. WEICK: Simply fairing from the nacelle which was already covered with MR. MONTEITH: That is correct, the center motor gave a gain of 5.1 [mph], NACA cowling clear to the wing without changing the location of it, and, inciden- but the three sets of NACA cowling added about 120 pounds to the ship, that is tally, that type of fairing was put on the ship and a total of increase of twelve miles almost one passenger. an hour was finally obtained. Also, the nacelle was moved in the model but it was CHAIRMAN MEAD: I wonder if Weick knows from Langley Field whether too much of a job to do that full scale, but on the model, the nacelle was moved to they have arrived at any conclusions about the relation of cowling and motors to the several positions above and below the wing (the only one which was moved was the wings? Isn’t that the reason it don’t work on Monty’s ship? nacelle with the complete NACA cowling) and it was found that by moving it far MR. WEICK: That is the reason that I ascribe to it, that is you have an interfer- enough down you could get away from fairing it, but it had to be an appreciable ence effect there between the nacelles and the wings, and so far in just about every distance below the wing before the interference was not noticeable. case I know of, which is three actual full scale tests, also some wind tunnel tests, The best results were obtained by putting the nacelle exactly in front of the the interference between a cowled nacelle and a wing nearby has been very great, in wing (indicating on the blackboard) with the top of the nacelle parallel to the top fact, great enough to usually nullify any possible gain due to the use of the cowling of the wing. on the engines. I will just take a small space here (illustrating). Supposing that is the wing (indi- 230 Chapter 3: The Design Revolution Document 3-11(a-b) 231 cating), and the normal nacelle came in about like this, and when it was in that I remember it, if you got the nacelle away from the wing with a gap down close to position we got no gain whatever, because the air would not flow through here, and one engine diameter you could get away from the interference, but that is more gap having large fillets all around, we got an appreciable gain, but by moving this up so than you can ordinarily get in an airplane. that the nacelle came about in this position (indicating) we got less drag than the MR. MONTEITH: You mean the space to be once the diameter? sum of the wing drag and the nacelle drag when each was isolated, that is, the total MR. WEICK: Supposing the diameter of that engine would be say 45 inches drag of this was less than the sum of the drag of the wing and the nacelle when they as it was on the J-5; if you put the top of the engine 45 inches below the wing, that were not together, so that it showed that we could get an increase in performance is what I mean. by changing the location of the nacelle and outing down the projected area of the MR. MONTEITH: On that question of Mr. Egtvedt’s here a moment ago, entire works if you had a smooth enough cowling. Of course, if you had an exposed Doctor Zahm made some tests four years ago on propellers in front of large wings. engine in that position you would spoil the lift over a good portion of your wing, The wings were much larger than the ones we are using however. and the Ford people with their first tri-motors ran into that problem, they had their MR. WEICK: Yes. Well, his tests, I think, were very good and they show one J-5 engines right in front of the wing, and they had such high landing speed they thing took, they show in a case like that, or in any case where the propeller is in front had to remove them. They took it at the time, it was due to the fact that the propel- of the wing, you have to put that appreciably in front of the wing or you get a very ler was in front of the wing. noticeable loss, and from his tests I got an idea that on an ordinary arrangement, we MR. McCARTHY: (Interrupting) You think if they cowled those in they would will say a nine-foot propeller, which was what we had on the Fokker that we were be all right? working with at the time, if we had put the nacelles up into the wing such as that MR.WEICK: Yes, because we did not get any decrease in lift by doing that, and then put the engines far enough ahead so that the propeller was approximately none whatever. We only did it at low angles of attack. I do not know what would 18 inches ahead of the leading edge of the wing instead of just, I think, about five happen at high angles of attack. or six inches as they were in the actual installation, there would have been an appre- MR. HAMILTON: High angles would be interesting as far as your landing ciable gain all the way through, and it looks to me off-hand as if that is the best speed is concerned, and that is where Ford had trouble. condition considering high speed. Now, what that particular condition would do to MR. WEICK: There are no tests giving reliable data in this Country. There that the lift at high angles or landing speed I do not know. have been a few small tests made in Germany, but they also did not give just exactly MR. HAMILTON: There is a Russian reproduction of the Ford tri-motor which what you want. I saw in London, where they have the engines out from the wing about better than Right at the present time, in fact the last thing I did when I left NACA was to two feet and it is not cowled. It has a Lynx engine and they claim they have tried outline a series of tests they are now working on and not in the tunnel yet, which the position of the Ford and also this position and the landing speed is no greater in consists of tests on a combination of a wing and a propeller in a nacelle, and the their position than with the Ford, than with the low wing position. However, their nacelle has forty different positions—approximately forty different positions with top speed is considerably improved. It seems to bear out that information due to the respect to the wing using the propeller as a tractor, and then there will also be a series fact they are I should say almost 30 inches ahead of the leading edge. of tests made using the propeller as a pusher that includes tests with the propeller CHAIRMAN MEAD: Well, it strikes me that one thing we would like to know entirely above the wing and entirely below the wing and three different positions as soon as possible is the results of these tests in NACA laboratories, with the pro- before and aft., so that it is right as near the wing as it could get clearance, and then peller turning. some positions for and aft., and in some cases like that where it is built right into the MR. WEICK: It is true, but you won’t get those for some time. wing and other cases where the nacelle is suspended above or below the wings. CHAIRMAN MEAD: I wondered if there is any chance of you or Chatfield; CHAIRMAN MEAD: Monty, are you going to have NACA cowling on the 200? somebody who knows them well, to get in there ahead of the rest of the gang? MR. MONTEITH: That is what we had planned originally, yes. MR. WEICK: Well, there is that possibility— CHAIRMAN MEAD: Based on what we know now, the trouble with NACA MR. HAMILTON: (Interrupting) They tell me, just with that idea in mind— cowling is heating, isn’t it? MR. WEICK: (Resuming) Actually, I just did go down and got all the dope MR. MONTEITH: I am still worried about that geared engine though. they have done on high tip speed so far. Those tests were started by me and they just CHAIRMAN MEAD: You mean in the 200 you are going to use geared finished them up after I left and I had no difficulty whatever in getting that stuff engine? because I know the fellows down there and because also I was in on the start of the MR. WEICK: With the cowled nacelle, which is the only thing we had there as thing and they realized that I had something to do with them down there. 232 Chapter 3: The Design Revolution Document 3-11(a-b) 233

Now it so happens that I was in on the start of this particular thing and will engines, the Wright J 4. Three of those were used, one in the nose of the fuselage, have no difficulty whatever getting that information by going there, but I can’t get the other two in the leading edge of the wing. This was the first development. It had it by getting them to write it to me. the engine, not in the nacelles beneath the wing, but right in the leading edge. CHAIRMAN MEAD: Then if we carry out our present scheme of getting such Pruden had designed the first trimotor, using such people as Tom Towle, John information to Chatfield he can broadcast it through the news letter and everyone Lee, Otto Koppen and McDonnell who were in the original Engineering Depart- will get it about as quickly as possible. ment with Stout. Koppen was in the Stout Engineering Department. He was not MR. WEICK: Only, any such information which comes out before it is pub- an independent outsider brought in to design the flivver plane alone. He was a part lished by NACA should be kept within the family because otherwise they, of course, of the original group, and I believe there were possibly ten to twelve fellows in the would not let us have anything further if they discovered that we were broadcasting group. When they came over, it was reduced to possibly seven because some of the it. It would get them in bad. men were not suited to aeronautical design. CHAIRMAN MEAD: You would not get any more? After I took over the Engineering Department, it had to be reorganized. I had as MR. WEICK: Yes, that is it, sure. my assistant, Tom Towle. He was an airplane designer and a very good one too. He CHAIRMAN MEAD: Then, as far as I can see at the moment we can’t help is now the president of the Church Company that makes baking soda. He married Monty very much except as to additional cooling in the engine. Church’s daughter, I believe. When I took over engineering, Mayo, of course, was the general manager. Stan Document 3-11(b), “The Reminiscences of Harold Hicks,” August 1951, Knauss had charge of the shop. Hoppin was the treasurer. Shortly after that, they transcript, pp. 61-69. Henry Ford Museum and Greenfield Village Research hired a man by the name of Rudolph Schroeder. He took over the flying end and Center, Dearborn, Mich. the airport operations. Harry Russell, I believe, worked for Stout originally as a mechanic. Manning Our position did not at first change when the Ford Motor Company took over was hired possibly a year or two after that. He was a young fellow that was a pilot Stout. At first, we ran along and did this overhaul work together with cleaning up and was hired into the Company after it had been fully taken over by the Ford other ends of the design on other small projects that I was doing at the time, includ- Motor Company. ing helping Horton on some stress analysis work. When I took over, they were producing the 2 AT planes. The 3 AT trimotored Along about the first of September, the “Shenandoah” dirigible cracked up in plane was being built and assembled. This first trimotor was powered by J 4 engines. Ohio. Mr. Ford wanted somebody to go there and give him a first-hand report. It was flown by Schroeder in possibly the late fall, maybe November of 1925. I can George Pruden, who was the Stout chief engineer and the man who is really respon- remember that there are photographs of that ship. Mr. Ford and Edsel were there. sible for the designing of the single engine plane, went down on the trip. I can remember when Schroeder brought it in. It landed awfully fast. It landed Being unfamiliar with the Ford Company way of doing things, he permitted a easily at eighty miles an hour. The reason for that was aerodynamic. We didn’t know picture of himself to be taken which was put on the front page of the Detroit Free how to cure it then because NACA had not developed their ring cowls to cover the Press with a caption showing Pruden writing his personal report of the crash to cylinders and prevent the disturbance from blanketing out the wing area. There was Henry Ford. That apparently didn’t suit Henry Ford very well so Pruden was fired. an inadequate wing area on the plane because the wing engines destroyed the air Then Mr. Ford came to me and asked me to take over the engineering of the flow behind them; that’s why it landed so fast. It was considered unsatisfactory. airplane. The only instructions that he made to me were very peculiar at that time. The wing section was one of the NACA sections, but it did not have a very He said that I was to get in there, and run it, and to keep Stout out of the design effective lift as it was convex on the bottom surface. That was the same type of wing room. He said that for the first time in his life, he had bought a lemon and he didn’t that was used on the single engines. want the world to know it. All the work that I did, together with any of the men, In his autobiography, Stout says that his original trimotor designed with the the credit would have to be given to Stout. engine in the leading edge was an advanced design over that with the pods under- He was very proud to think that he was a good businessman. Apparently Mayo neath the wings. He says that the only reason he changed the design was because of had led him into something, not fully realizing that Stout was a promoter and not the fact that the officials of the Ford Motor Company thought that the plane landed a practical engineer. too fast. In lowering the engine and putting it in a pod beneath the wing, they were At the time when Pruden was released from the Company, they had under con- actually causing a decrease in the performance of the plane. struction a three engine plane which was known as the 3 AT job. That had air cooled That is probably so, but at the same time the plane with the engines in the 234 Chapter 3: The Design Revolution leading edge of the wing without any NACA cowl rings would have been entirely unsatisfactory. It landed too fast, and the top speed of the plane wasn’t much either. As I remember, the top speed was only about 110 miles an hour. So with a top speed of 110 and landing speed of 80, you didn’t have any range at all. The take off speed was around 80 miles an hour too. Schroeder, according to Aviation News, is in Chicago. This was about six months ago. He is writing his autobiography. However, the very fact that the plant burned down, in my opinion, proved a benefit to the Ford Motor Company and to aviation in general. Plans were immediately made to start a much better factory, one which was built quite quickly. It is now the body building and garage. At least the west half of it was erected very quickly after the fire. After the fire Mr. Ford wanted to go on. The fire had occurred so early in the Ford experience of producing airplanes that that is probably the one reason why he built it up rather than to cross it off the books entirely. Document 3-12(a-h) 235

Document 3-12(a-h)

(a) Harlan D. Fowler to Rueben H. Fleet, 8 January 1929, Folder “Fowler Flap,” copy in San Diego Aerospace Museum, CA.

(b) Consolidated Aircraft Corporation to Harlan D. Fowler, 21 January 1929, Folder “Fowler Flap,” copy in San Diego Aerospace Museum, CA.

(d) Harlan D. Fowler to William B. Mayo, 15 August 1932, copy in Accession 18, Box 96, Folder “Fowler Variable Wing,” Henry Ford Museum and Greenfield Village Research Center, Dearborn, Mich.

(e) H.A. Hicks to Harlan D. Fowler, 10 September 1932, copy in Accession 18, Box 96, Folder “Fowler Variable Wing,” Henry Ford Museum and Greenfield Village Research Center, Dearborn, Mich.

(f) T.P. Wright, “The Application of Slots and Flaps to Airplane Wings in America,” U.S. Air Services (September 1933): 29-31.

(g) Robert C. Platt, “Aerodynamic Characteristics of a Wing with Fowler Flaps including Flap Loads, Downwash, and Calculated Effect on Take-off,” NACATechnical Report 534 (Washington, 1935). 236 Chapter 3: The Design Revolution Document 3-12(a-h) 237

(h) Harlan D. Fowler, “Aerodynamic Characteristics of the no longer working as an engineer but out of the aeronautics world as a salesman Fowler Wing,” Aero Digest 29 (September 1936): 46-50. in California. In 1933, he traveled to Baltimore in hope of convincing Glenn L. Martin of the value of his flap. The fact that the NACA had been experimenting with the device was a decisive factor in how Martin responded. He gave Martin a Harlan D. Fowler’s design of a wing flap in 1924 is one of the least known yet job and put him to work devising flaps for several new Martin aircraft. One of these classic stories of invention in the history of American aeronautics. An employee of became the Martin 146 bomber of 1935. Although it was never produced, the air- the army’s Engineering Division facility at McCook Field in Ohio, Fowler became plane did employ Fowler flaps. interested in the concept of a variable-area wing late in World War I. For the next six By the mid-1930s, the U.S. aeronautics community as a whole had grown years, he developed several unsuccessful prototypes until he came up with the idea extremely interested in the promise of high-lift devices. Airplanes were flying so fast of combining a slotted flap similar to the Junkers auxiliary wing with an increase and growing so large that great improvements in lifting capacity and much higher in wing area and camber. In 1929, this notion led him to design what came to be wing loadings were becoming an absolute necessity. Lockheed incorporated Fowler known as the “Fowler flap,” a small airfoil housed in the trailing edge of the wing. flaps into its new twin-engine airliner of 1937, the Lockheed 14. It was the first When needed, the airfoil could be extended backward until there was a slot between production-line airplane to use Fowler flaps. (A few German aircraft with Fowler- it and the main wing, and then hinged downward. Possessing a much heavier mech- type flaps had appeared earlier.) After that, a growing number of planes started to anism than a slotted or split flap, the Fowler flap made up for the weight with the use slotted flaps of various kinds. Fowler became one of the most recognized names increase in wing area and greater lift, which improved both takeoff and landing per- in all aeronautical engineering. Even today, his name is still associated with a generic formance. Thus, the basic innovation was to increase lift by combining two effects: brand of slotted flaps used on many aircraft. increasing the camber or curvature of a wing by deflecting the flap downward, and In the string of documents below, one returns to the early 1930s to see how the increasing the wing’s surface area by mechanically extending it out downstream of American aeronautics community first became seriously interested in the Fowler the wing. Used together in such a way, the combination would greatly shorten take- slotted flap. off runs, lower landing speeds, and increase rates of climb. The army did not support Fowler’s work, which was okay with Fowler as he had always intended his flap as a private venture. He began turning his idea into a practical mechanism in 1926. Working in his spare time and with his own funds, he constructed wings of his own design for various biplanes through the late 1920s. Flight tests revealed that landing speed decreased by two miles per hour and overall top speed increased by ten miles per hour. During this period, Fowler claimed that his design might deliver a whopping 90 percent boost in lift, which few people were prepared to believe. No one paid attention to the “Fowler Variable Wing” until the NACA started taking a look at it in 1932. Wind tunnel tests at Langley during that year proved that his design did in fact contribute to the overall performance of an airplane. The NACA engineer most involved in testing the Fowler flap was Fred E. Weick, spear- head of the NACA’s cowling and engine nacelle placement programs. Weick felt that Fowler’s design merited at least some limited wind tunnel testing. In the first report on the Fowler flap published by the NACA in May 1932, Weick reported on “wind tunnel tests of the Fowler variable-area wing” (Technical Note 419). A year later, another report appeared, “Wind Tunnel Tests on model wing with Fowler flap and specially developed leading-edge slot” (Technical Note 459, May 1933), co- authored by Weick and Robert C. Platt. The NACA’s interest and its promising technical evaluations gave Fowler’s invention its first true stamp of credibility—one that he really needed as he was 238 Chapter 3: The Design Revolution Document 3-12(a-h) 239

Document 3-12(a), Harlan D. Fowler to Rueben H. Fleet, 8 January 1929, FOWLER VARIABLE AREA WING Folder “Fowler Flap,” copy in San Diego Aerospace Museum, CA. The demand of aircraft operators for improved performance and load carrying capacity to meet the requirements of the public is bringing about a condition which FOWLER AIRPLANE WINGS, INC. will compel consideration of advanced wing design. Primarily, an aircraft must have Manufacturer of Variable Area Wings sufficient wing lift to provide safe landings and take-offs. This restriction has a NEW BRUNSWICK, N. J. direct influence on high speed and load capacity and is an economic loss. July 8, 1929. The Fowler Variable Area Wing provides for this condition. The powerful com- Mr. R. H. Fleet, President, bination of the advantages of change in camber, area, angle of incidence and aspect Fleet Aircraft, Inc. ratio, simultaneously, has resulted in a wing design of tremendous value to aircraft 2050 Elmwood Ave., operators. Buffalo, N. Y. As the result of further research and development of the Fowler Wing substantial improvements in performance are obtained. The following comparative data Dear Sir: relative to the original PA-3, OXX-6 100 H.P. is interesting: The development of the Fowler Variable Area Wing is approaching the stage of practical utility after being successfully used on several types of planes. Original Fowler Wing A standard type of construction and wing area is now being prepared with the purpose of applying it to the larger majority of types of airplanes now in use giving Performance PA-3 no. 1 no. 2 them very substantial improvements in all around performance, that is, lower land- Low speed, normal, M.P. H. 46.2 58.2 57.0 ing speed, higher top speed and good climb. Low speed extended M. P. H. -- 44.5 41.0 We are approaching you as a manufacturer of high grade airplanes with a pro- High speed, M. P. H. 91.0 101.0 104.0 posal involving the use of our wing in place of your standard product for a cost Cruising 77.0 84.0 87.0 within the range of your own factory cost wings. Rare of climb, normal, ft. / min. 515 430 570 If this proposal meets with your approval and you will assure us of your inten- Total weight of plane 1936 1990 2000 tion to contract for a limited number of wings, then we will install our wing on your Wing area, normal, sq. ft. 338 136 145 plane and obtain complete comparative performance to be used by your company Wing area, extended, sq. ft. -- 166 186 for sales purposes, and cooperate with you in giving demonstrations. Wt./ sq. ft. normal 5.72 14.65 13.80 This arrangement will give your products a distinct step in advance of the con- Wt./ sq. ft. extended - 12.00 10.75 ventional plane and consequently greatly influence your sales prospects. Speed range 1.97 2.16 2.42 We would appreciate learning how this procedure appeals to you and we will be pleased to submit such information as you may desire. The above performances are based on the best propeller design suitable for that type of plane. Performances for the PA-3 and Fouler Wing No. 1 are based on actual Respectfully yours, flight tests over a one mile course and are the average of several runs. Knowing the high speed of Fowler Wing No. l the parasite resistance was obtained after subtract- President, Fowler Airplane Wings, Inc. ing the wing resistance. With this parasite resistance still used and with the new Fowler Wing No. 2 substituted the above performance was obtained. Reference to the above performance reveals a number of possible applications of the Fowler Wing No. 2 as applied to this particular low powered type of plane: (a) As between the biplane the low speed is reduced almost 5 miles per hour. Between the usual monoplane (normal area) it is 16 miles per hour less in low speed when the extension area is used. Note the difference in wing area. (b) The high speed of the normal. wing is about 13 miles per hour faster than for the biplane. 240 Chapter 3: The Design Revolution Document 3-12(a-h) 241

(c) The load carrying capacity of the extended wing per square foot area is over Document 3-12(b), Consolidated Aircraft Corporation to Harlan D. Fowler, 100% greater than for the biplane, and with less landing speed. 21 January 1929, Folder “Fowler Flap,” copy in San Diego Aerospace Museum, (d) Speed range for the Fowler Wing is increased almost 23 per cent over the CA. biplane. (e) Rate of climb for the Fowler Wing is of a very high order considering its other remarkable features. (f) The Fowler Wing is only 25 pounds heavier than the biplane with a January 21, 1929 strength factor 54 per cent greater than the original PA-3. The evidence presented by these tests indicates possibilities such as: Mr. Harland D. Fowler (a) Pay load increased 100% for equal performances and power. The Miller Corporation (b) Same pay load and performance for about once-third less power. New Brunswick Airport (c) A reduction in wing area 50%. For large planes this is of extreme New Brunswick, N.J. importance. Letter Patents have been granted for the design of this device. Dear Sir: Fowler Wing No. 2 is of such area that it can be adopted to 75 percent of the commercial types of planes now in use, giving them an all around improved per- Could you send us information on the typical detail design of your variable area formance. This will be possible on planes using power up to 225 H. P. and a total wing? Also, tell us what the span of the PA-3 biplane which you used was, both as a plane weight of 3000 lbs. Fowler Wing No. 2 is to be put in production and made biplane and as a monoplane with normal and standard wings. available to the industry. The Fowler Wing is also available to solve any special problems which it may Very truly yours, be desired to accomplish. CONSOLIDATED AIRCRAFT CORPORATION Fowler Airplane Wings, Inc. New Brunswick, New Jersey. Asst. to General Manager 242 Chapter 3: The Design Revolution Document 3-12(a-h) 243

Document 3-12(c), Harlan D. Fowler, “Variable Lift,” Western Flying 10 Variable angle of attack; (4) A variable camber, and (5) The slotted wing. (November 1931): 31-33. The lift of the normal or closed wing is increased about 100 per cent by the combined use of these relations. Each feature contributes to the increased lift in the following proportions: Variable camber, 43 per cent; variable area, 28 per cent; November, 1931 recessed camber, 7 per cent; and slots, 22 per cent. And it is all accomplished by the one simple expedient of extending the auxiliary area wing downwardly and rear- VARIABLE LIFT wardly, which rapidly brings about the simultaneous characteristics of each desirable BY: HARLAN D. FOWLER relationship. In spite of the achievements of the Guggenheim Safe Aircraft Competition, little progress seems to have been made toward the solution of the problem of increasing Wind Tunnel Tests the lift of wings without sacrificing top speed. The following article is presented with the hope that it will stimulate more thought on the subject. -The Editors. Figures 1 and 2 give the characteristics for the normal and extended area. The Time consumed in taking off and landing is a matter of a few minutes, whereas tests were conducted at the New York University. The maximum Ky normal is straightaway flying takes hours. Simply because sufficient lift must be available for .0036 and extended .00556. The normal area is 63 sq. in. and extended 80 sq. in. this landing and take-off, we afflict ourselves with the use of oversized wings. When Viewed on the assumption that the normal wing characteristics are influenced by once in the air, the smallest of wing, consistent with controllability, can be used these alterations by virtue of being self-contained, it will have the equivalent maxi- safely. mum Ky of Why, then, do we continue to be so far removed from the most important and × far reaching development of variable lift wings? As far back as 1910, the subject .00556 80/63 = .00705 of utilizing variable lift wings was frequently discussed by engineers. The methods suggested were: or (a) variable camber, − (b) variable area, .00705/.0036 1 = 0.96 increase (c) variable angle of attack, and later (d) slotted wings. Full flight speed trials at near stall gave a maximum Ky of .0073 as referred to Many laboratory experiments have been made on airfoils with the camber the normal area—the highest lift obtained from any airfoil. adjustable from a streamline section to that of a heavily convexed section, with This increase in lift was obtained without impairment of aileron effectiveness, promises of a wide variation in aerodynamic characteristics. Variable area has pre- only 70 per cent of the wing span being utilized for the auxiliary wing. sented a mechanical problem of the first order. The variable angle of attack offers the least, if any, improvement, although several airplanes have been tried out using this idea. Perhaps the most successful devices have been the various forms of slotted wings. All four methods have been considered extensively in published articles. However, it is unfortunately true that the benefits obtained from each type alone have not justified its continued development.

The Fowler Wing Principle It was long recognized by the writer that if the best features of each one could be incorporated into a simultaneous and coordinated alteration of the wing, the fullest advantage would be utilized. This was easier “thought of” than practically solved. Fundamentally and foremost, the basic structure of the wing must not be interfered with. Hence, the conception of the Fowler Variable Area Wing, which incorporates the following points: (1) A structurally sound basic wing; (2) An increase in area; (3) 244 Chapter 3: The Design Revolution Document 3-12(a-h) 245

Full Flight Tests Performance tests were made with a Pitcairn PA-3 which was originally a biplane of 338 sq. ft. An OXX-6 engine of 100 h.p. and the same wood propeller was used throughout the trials without alterations or overhauling . The following data were obtained: O original Fowler Wing PA-3 no. 1 no. 2

Low speed, normal, M. P. H. 46.2 58.2 57.0 Low speed, extended, M. P. H. --- 44.5 41.0 High speed, M. P. H. 91.0 101.0 104.0 Cruising speed, M. P. H. 77.0 84.0 87.0 Rate of climb, ft./min. 515 430 570 Take-off time, seconds 10 11 10 Landing time, seconds 9 11 9 Total weight of plane, lbs. 1936 1990 2000 Wing area, normal, sq. ft. 338 136 143 Wing area, extended, sq. ft. --- 166 181 Wt. per sq. ft., normal 5.72 14.65 13.8 Wt. per sq. ft., extended --- 12.00 10.75 Speed range 1.97 2.16 2 42 Design load factor of wing 5.50 8.00 10.00 Weight of wings, complete, lbs. 377 395 409

Performances for the PA-3 and Fowler Wing No. 1 are based on flight tests over a one-mile course and are the averages of several runs. Wing No. 2 is an entire new design with a special normal airfoil and aspect ratio of 9. It is designed for a gross-weight of 2850 pounds with a load factor of 7, which for-the 2000-pound plane is equivalent to a load factor of 10. It will be noted that in spite of using an extended area of less than half that of the biplane, the stall speed was nearly 2 m.p.h. lower. The stall speed for the normal wing monoplane is 58.2 m.p.h. and represents the typical monoplane. By extending the area, this speed was reduced to 44.5 m.p.h. Thus, the average monoplane can have its stall speed reduced not less than 14 m.p.h. The high speed was increased from 91 m.p.h. for the biplane to 101 m.p.h. for the normal area Wing No. 1. For the original biplane to have attained 101 m.p.h., at least 136 horsepower would be necessary, resulting in a heavier engine and increased fuel capacity, or an increase in weight of about 185 pounds, using a water-cooled engine. This in turn would raise the stall speed from 46.2 to 48.7 m.p.h. The most important fact is this—the additional cost of a new engine at about $2,000 far exceeds the cost of a set of variable wings. 246 Chapter 3: The Design Revolution Document 3-12(a-h) 247

Weight Comparison and straightforward, being governed entirely by conventional design practice. No Wing No. 2 has been designed according to the strength requirements of the unusual or tricky supporting members nor articular parts are used in the wing. Department of Commerce and approved for installation on planes with gross weight For extremely high speed purposes, it is necessary to go into high landing speeds. up to 2850 pounds and up to 225 h.p. The actual weight of the wing is itemized The Schneider Cup Race planes land at speeds varying from 100 to 125 m.p.h. as follows: Let us assume that with a high speed airfoil, the wing loading is 30 pounds per sq. ft. Substituting the variable area wing, the extended wing loading would be 60 Main wing, with fittings 242.501bs. pounds per sq. ft. at 100 m.p.h. stall speed and normal area is 76 pounds per sq. ft. Ailerons 17.001bs. at 145 m.p.h. stall speed. This is a reduction of 45 m.p.h. It is acknowledged that Five gallons gas tank 8.50 lbs. the wing loadings are very high, but the wing could be of a cantilever type and very Main wing, complete 268. small in size. The high speed would be greatly improved. Extension surface 36.00 When landing with the wing extended, the glide is steep and the run on the Extension surface rails, etc. 24.00 ground short. Extension surface controls 6.00 Extension assembly, complete 66. Wing Reduces Need for Increased Power Fowler Wing No. 2, complete 334. If increased high speed is desired, it is usually the practice to put in a higher power engine. This will not be necessary if the Fowler wing is used. Struts and wires for installation to plane weigh about 75 pounds, giving a total Let us consider a pure cantilever monoplane with a 420-h.p. engine, wing area weight of 409 pounds. 265 sq. ft. and a gross weight of 4500 pounds. The pay load is 600 pounds without This wing can be designed for a closed cabin without special difficulties. including 150 gallons of fuel. Stall speed is 66 m.p.h. For comparison with the conventional wings, the following typical data are given: Installation of a 525-horsepower engine would only increase the high speed of this plane from 200 to 215 m.p.h. It is an unfortunate fact that in changing to Rated gross Area Weight Stall load higher power, the expected increase in speed is rarely obtained. The average loss is H. P. Weight sq. ft. Wings Speed Factor about one-third of the theoretical increase. This is largely attributed to the increase in cylinder number and diameter, overall engine diameter, and loss of propulsive Fowler Wing 225 2850 181 409 53.0 7.00 efficiency. Therefore, it is very probable that the high speed of 215 m.p.h. for the “A” biplane 180 2877 351 425 49.0 7.00 525-h.p. engine may be more like 210 m.p.h. or even less, if available data are “B” biplane 165 2702 289 450 54.0 7.00 trustworthy. The added cost of the larger engine is about $600. The larger gas tank to receive Note—Even though Wing No. 2 is lighter than the conventional wings, no 30 gallons more fuel, propeller, etc., would add at least $100 more, or a total of $700. attempt was made to resort to expensive light-weight detail construction. With such A variable wing would cost about 15 per cent or $375 less than the original an excess of lift available, this is unnecessary. However, with design refinement, at wing. The total net initial saving would be $1,075. The operating savings on gas least 10 per cent further saving in weight is possible. A and B are well-known air- consumption, depreciation of engine, and maintenance would also be considerable. planes. Climb and Ceiling Application to Cantilever Wings Numerous tests with the extension in various locations between all-in and all- As shown before, for the same stall speed, the area may be reduced one-half by out shore a very definite gain in the rate-of-climb when set about halfway out. This using the variable area wing. Since the extension feature can be easily incorporated is, then, the best setting for take-off and climb. in a cantilever wing, the span can be reduced 25 per cent for the same aspect ratio. With the fully extended wing, the rate of climb is lower, as is to be expected, in Since the gross weight is about the same, the cantilever bending moment is corre- view of the lower L/D. Incidentally, for the same reason, the high speed was found spondingly reduced. A stiffer and cheaper structure would result from this reduction to be about 79 m.p.h. This condition is in reality a decided safety measure in that in size. it would be impossible to maneuver sharply. It thus prevents undue strain on the From the standpoint of strength, the determination of structural sizes is simple extended surface and supports. 248 Chapter 3: The Design Revolution Document 3-12(a-h) 249

The planes on which the Fowler wing was used had fixed stabilizers, and these were not readjusted at any time. The extension can be operated by the pilot while flying and set at any position desired. The position of the control stick is changed forward when extending the wing, but little change of balance is noted for the two conditions of all-in or all-out. The plane had adequate control at and beyond stalled position.

Design Features The normal wing, which is of two halves joined at the center, is of conventional These rails are attachable in up or down movements by eccentrics located at the construction, although employing materials in a different manner. The spars are of main fitting attachments so as to provide clearance between the auxiliary wing and solid spruce. The ribs are of 3/8 and 7/16-inch spruce. The nose is of solid balsa the under surface of the main wing when it is closed. wood. (Fig. 4.) No internal drag bracing is used because the covering is of plywood Operation of the extension is by means of a continuous cable wrapping around which is glued and nailed directly to the spars and ribs. a drum. To this drum is secured the control shaft and handle reaching down to the The auxiliary wing has a single spar to which are secured the solid ribs and balsa pilot who can operate it at his convenience. All control cables, including those for wood nose piece. The whole is plywood-covered. This small wing fits snugly under the ailerons, are led along the open face of the rear spar. Nothing is concealed. the trailing portion of the main wing. (Fig.4.) Commercial Applications Due to the high lift of the Fowler wing, it can be applied as follows: (1) Light planes, stall speed 25 m.p.h.; (2) cargo planes, actual payload, 7 pounds per h.p.; (3) optional equipment, substitute for conventional wings, and (4) reduced initial cost of planes and better performance.

Figure 5 is a drawing of the complete combination, showing the supporting rail, trolley and relation of the large and small airfoils. The four supporting rails, which are 3/16 inch thick by 3 1/8 inches deep, are of steel, but may be of duralumin. 250 Chapter 3: The Design Revolution Document 3-12(a-h) 251

Document 3-12(d), Harlan D. Fowler to William B. Mayo, Document 3-12(e), H.A. Hicks to Harlan D. Fowler, 10 September 1932, 15 August 1932, copy in Accession 18, Box 96, Folder “Fowler Variable Wing,” copy in Accession 18, Box 96, Folder “Fowler Variable Wing,” Henry Ford Museum and Greenfield Village Research Center, Dearborn, Mich. Henry Ford Museum and Greenfield Village Research Center, Dearborn, Mich.

615 Fifteenth Avenue, San Francisco, Cal. Sept. 10, 1932. July 19, 1932 Mr. Harlan D. Fowler, 120 Webster Street, San Francisco, Mr. Wm. B. Mayo, Cal. Ford Motor Company, Airplane Division, Dear Sir: Dearborn, Mich. In reply to your letter of August 15th concerning your variable area wing, we inform you that this company would not be interested in purchasing manufacturing Dear Mr. Mayo: rights at this time. With reference to yours of June 28th, I am taking the liberty of sending a copy of the N.A.C.A. report No. 419 (on a loan) so that a study can be made of the Fowler Variable Area Wing. Yours very truly,

We have found Position No. 6 to be an all around ideal arrangement and to FORD MOTOR COMPANY which the enclosed blueprint of the characteristic curves applies. (Airplane Division)

It may be to your interest to learn that several other large companies are consid- H.A. Hicks, ering the use of the Fowler Wing. If your research is convincing I shall be pleased to Aircraft Engineering. consider a mutual arrangement.

Very truly yours, Fowler Airplane Wings, Inc. 252 Chapter 3: The Design Revolution Document 3-12(a-h) 253

Document 3-12(f), T.P. Wright, “The Application of Slots and Flaps to Airplane The flap is a device which has for its essential object the increasing of the camber Wings in America,” U.S. Air Services (September 1933): 29-31. of the basic wing section to which it is applied. In its simplest form the flap consists of a hinged portion of the trailing edge of the wing, which, when deflected down September, 1933 ward, obviously increases the wing’s camber. One variation of this simplest form of flap is accomplished by changing the form of the leading edge of the flap, to create The Application of Slots and Flaps to Airplane Wings in America a slot at that point, which, as in the case of the slot at the leading edge of the main T. P. WRIGHT airfoil above described, gives the beneficial effect of smoothing out the flow over the Vice President and Director of Engineering flap when deflected. Curtiss Aeroplane and Motor Company, Inc. Another form of flap is described as the split trailing edge type whereby only the lower portion of the trailing edge is deflected downward in operation, rather than It is the object of this article, within the space limitations set, to present in the whole trailing edge, as in the case of the simple-hinged flap. In a further varia- general terms the objective sought in incorporating slots and flaps on airplane wings tion of this type, this lower portion is moved aft at the same time that it is deflected and to indicate briefly the application which has been made in this country. downward, thus to a certain extent increasing the wing area in addition to increasing its camber. A still further variation in flap design is incorporated in the Fowler Airplane Characteristics type in which the lower split flap section is made up into the form of an airfoil and For the purpose of this article it appears desirable to segregate the qualities of is moved aft through its chord length in addition to its downward angular deflec- the airplane which are affected by these devices into two main groups. These are tion and is so shaped that when in this final operating position a slot is formed at qualities affecting safety and qualities affecting performance. its leading edge. Under the heading SAFETY the functions which are affected by changes in The increase in maximum lift due to the type of flaps above described varies wing characteristics are landing speed, susceptibility to spinning, and controllability, but in all cases is substantial. An extensive series of tests has been conducted by the particularly beyond the stall. Other things being equal, it appears obvious that the National Advisory Committee for Aeronautics, and the results tabulated, to show airplane with the lower landing speed, the airplane having freedom from incipient the effect on airfoil characteristics of all of these types of flap and of the flap in spinning tendencies, and the airplane which maintains lateral control beyond stall- combination with leading edge slots. It is shown in these tests that, based on basic ing speed will be the safer craft. wing area, lift increases of more than 100% are readily obtainable with accompany- Items of performance which are affected by changes in wing characteristics are ing speed range increases of more than 75%. The possibilities of such tremendous high speed, rate of climb and ceiling, climbing angle, and angle of descent. Higher improvements in airfoil characteristics are obvious and when airplanes designed and higher speeds are desired each year as aviation progresses so that the advantages with them have other characteristics properly correlated, it is believed that without of any device tending to increase speed are obvious. Rate of climb and ceiling are doubt they will find a substantial application in the majority of types of airplanes. only of relative importance, the degree of importance depending, of course, upon the particular type of airplane under consideration. The last two functions, angle of Improvement in Airplane Characteristics Obtainable by climb and angle of descent, are becoming recognized as being items of major impor Use of Slots and Flaps tance more and more as the external cleanness of airplanes increases. Let us now note the possibilities which these devices hold for improving our fundamental airplane qualities, namely, safety and performance. Although the slot Description of the Slot and Flap can contribute slightly to the lowering of the landing speeds, its chief utility lies in Let us now briefly define the slot and flap. The slotted wing is one which has its effect in preventing spins and in permitting lateral control at high angles. It has an auxiliary airfoil at its leading edge, either fixed or movable, so that in either case, one other possibility when interconnected with the flap, namely, the supplying of in its extended position, a slot is formed which induces a smooth flow over the the necessary force for operating the flap, thus relieving the pilot of this operation. wing at higher angles of attack than are obtainable with the omission of the device. A properly designed auxiliary airfoil when in the unslotted position will increase It is important to note that the slotted wing retains the same slope of lift curve as the wing drag inappreciably and therefore not affect the high speed of the airplane maintains with the unslotted wing. The function of the slot is to increase the angle adversely. The high speed of the airplane can, however, be increased for a given at which burbling occurs. Therefore, if other airplane characteristics are arranged to landing speed by the use of the flap with its increased lift coefficient and resulting conform with the use of a higher lift angle, then the maximum lift of the airplane is decrease in wing area, although the degree to which this application can be used substantially increased by the use of the slot. is contingent upon the allowable resultant characteristics which are injured by the 254 Chapter 3: The Design Revolution Document 3-12(a-h) 255

application, namely, rate of climb, ceiling, and angle of descent (perhaps better It is believed that in the Commercial field an even wider application is probable defined as sinking speed). However, increase in sinking speed or angle of descent to both because of the increased importance of safety and because of the lesser impor- a limited extent is a decided advantage and one which is becoming more and more tance, in most cases at least, of the characteristics which are admittedly injured by essential in preventing floating during landing in modern airplanes having ever- the use of flaps or flaps in conjunction with slots. increasing fineness. The improved angle of climb obtainable by the use of flaps is also a quality of importance in airplanes of modern design. Development in this Country In summary, therefore, we find that, with the slot, improved safety is the chief With the above discussion, covering a description of the devices we are con contribution with some possibility of increasing lift and therefore speed range, pro- sidering and of the airplane characteristics which they alter, thereby indicating their vided the airplane is otherwise designed to permit of the realization of the high possible usefulness in the various types of airplane being developed, before us, let angle of attack necessary. With the flap a very definite increase in speed range is us now see what application has been made in this country. The basic scientific obtainable together with improved takeoff and landing characteristics. development of the slotted wing was initiated in England and Germany. Apparently We thus see that it is possible to improve certain performance characteristics the basic idea of improving airflow by means of the Venturi action of a slot was of the airplane by the use of slots and flaps. If the basic considerations determining coincidentally and independently worked out by Messrs. Lachmann and Handley the design of a particular type coincide with the characteristics which are improved Page. Credit for the major engineering development with reference to specific and by these devices, it then appears that their use is justified. There are certain other practical devices is due to Handley Page in England. Important note of this develop- types of airplanes wherein the basic characteristics governing the design are injured ment was first taken in this country in 1927 when the Navy Bureau of Aeronautics by the installation of slots and flaps, thereby making their adoption of no value or sent representatives to England to study the progress made and later negotiated cer- of questionable value. tain license contracts with Handley Page. Shortly thereafter several applications of the slot were made on Naval Aircraft, including Curtiss F7C-1 single seater fighter, Types of Airplanes a Consolidated training plane, and a Vought observation plane. This application There are two major classifications into which airplanes may be placed, namely, consisted of the installation of slots at the outer portion of the wing, forward of the MIILITARY and COMMERCIAL. Within each group there are subdivisions such aileron, only. as training, fighting, attack, observation, and load carriers, including bombers and The first commercial application seen in this country was on the Moth airplane, transports, in the former; and training, utility, private owner and transport, includ- constructed and sold here in considerable numbers under British license in 1928 ing mail and passenger, in the latter. Each type demands a special combination of and 1929. In this case also the slot at the wing tip only was used. characteristics to permit of the best fulfillment of its functions . It will probably be The first combination of slots along the whole span of the wing, with flaps agreed that the function of safety will be of less importance in military airplanes also on the whole span (floating wing tip ailerons being used in this case for lateral than commercial, and within the list of commercial types will be of less importance control) appeared on the Curtiss Tanager, winner of the Guggenheim Safe Aircraft in the mail plane and utility plane than in the case of the private owner plane and Competition in 1930. It should be noted that the plane most nearly approximat- passenger transport. This statement therefore in itself simplifies the study neces- ing the Tanager performance in the competition also employed slots and hinged sary in deciding upon the use of slots alone on a given design. The decision on the flaps, although in this case conventional ailerons were used. This was the Hand- application of flaps, however, is more difficult, as there are few classes of plane in ley Page Gugnunc. It is believed significant that the winner and runner-up in this either broad subdivision which are not susceptible of some improvement by the use competition were both equipped with slots and flaps, especially as it is considered of flaps, or flaps in combination with slots. that the tests set up in the competition correctly represented airplane characteristics essential to safety. Subsequent developed improvements in general airplane design Specific Applications could equally well be applied to the Tanager. The Tanager development was the In the case of military planes it appears that the attack type, requiring high result of intensive technical research in the Garden City laboratory of the Curtiss speed with a lesser degree of climb and ceiling than other types, represents one pos- Aeroplane and Motor Company during the period 1927 to 1929. The airplane was sible application. This applies particularly to Army planes. An additional problem is delivered to the Guggenheim Committee for tests in October, 1929. It is believed involved in the case of the Navy where low landing speed for carrier operation is of that this theoretical wind tunnel and full flight research was the most extensive far greater importance than in the case of the Army. Therefore in the Navy it appears carried out to that date in this country. The data determined from this study have logical that a wider application may be found desirable. been substantially augmented by the results of the very fine program of testing sub- 256 Chapter 3: The Design Revolution Document 3-12(a-h) 257

sequently undertaken and accomplished by the National Advisory Committee for Document 3-12(g), Robert C. Platt, “Aerodynamic Characteristics of a Wing Aeronautics. with Fowler Flaps including Flap Loads, Downwash, and Calculated Effect Although some work was carried out in 1930 on the application of slots and on Take-off,” NACA Technical Report 534 (Washington, 1935). flaps to commercial, private owner class airplanes, the development was curtailed and then abandoned because of the need for economy at the start of the depression. REPORT No. 534 Undoubtedly during the last three years a great number of successful applications AERODYNAMIC CHARACTERISTICS OF A WING WITH FOWLER FLAPS would have been made but for the above economic conditions. INCLUDING FLAP LOADS, DOWNWASH, AND CALCULATED EFFECT The next application appearing in this country, and the first one applying to ON TAKE-OFF a military airplane, was the use of both slots and flaps on the Curtiss XA-8 developed as an attack airplane for the Army Air Corps. The ship is the low, thin-wing, By ROBERT C. PLATT externally braced monoplane type, carrying a useful load about 50% greater than maintains for other classifications of two-seater military plane. The requirement of SUMMARY high sea level speed is of major importance in this class. This airplane flew in the This report presents the results of an investigation in the N.A.C.A. 7- by 10- summer of 1931, exceeding requirements substantially . The excelling performance foot wind tunnel of a wing in combination with each of three sizes of Fowler flap. of this plane represented the first large step in high speed increase of Army Air The purpose of the investigation was to determine the aerodynamic characteristics Corps planes since the War as it roughly stepped up speeds from the 150 m.p.h. to as affected by flap chord and position, the air loads on the flaps, and the effect of the 200 m.p.h. bracket. This ship, at the time of its appearance, led the Military Field, flaps on the downwash. The flap position for maximum lift; polars for arrangements when judging excellence on the criterion of load carrying times speed per given considered favorable for take-off; and complete lift, drag, and pitching-moment horsepower. Subsequently 13 similar ships were constructed and service tested, fol- characteristics for selected optimum arrangements were determined. A Clark Y wing lowed by procurement in a production order of 46 now underway. model was tested with 20 percent c, 30 percent c, and 40 percent c Fowler flaps of Other applications to military developments, both in the Army and Navy, have Clark Y section. Certain been made and are being made, the results of which can only be described and their additional data from value judged after service testing is completed and official release of information is earlier tests on a similar available. model equipped with Commercially, where from basic considerations application should be the most the 40 percent c Clark general, slots and flaps have been decidedly neglected at least until very recently, Y flap are included for where several applications of the flap have appeared. This has undoubtedly been due comparison. Results of to the economic conditions of the country and to the absence of properly charted calculations made to information on characteristics of airfoils with slots and flaps, now fortunately avail- find the effect of the able in published National Advisory Committee reports. Fowler flap on take- At the recent conference at the Langley Field Memorial Laboratory of the off, based on data from National Advisory Committee for Aeronautics considerable stress was laid on the these tests, are included possibilities in the use of various types of flap and of these types in combination in an appendix. with slots. The interest in this country, however, in slots continues meager in spite The maximum lift coefficient obtainable, based on original wing area, had a of the quite general adoption of these devices in various forms on a number of nearly linear increase with flap chord up to 40 percent, but the maximum lift force planes in several European countries. One other consideration which has undoubt- per unit of total area increased very little beyond the value obtained with the 30 per- edly retarded general interest in these devices is the fear of additional complexity cent c flap. The maximum load on the flap occurred very nearly at the maximum lift and weight increase due to their use. The importance of these latter features is fully of the wing-flap combination and was nearly 1½ times the load that would result realized, but nevertheless it is believed that if, as is maintained, the aerodynamic from uniform distribution of the total load over the total area. In general, the flap improvements obtainable by their adoption are substantial, and assuming intelli- appeared to carry a large proportion of the additional lift caused by its presence and gent application in each specific case, the obstacles of complication and weight will to have its center of pressure much nearer the leading edge than it would normally be surmounted and a wider use of slots and flaps made in the future. be in free air. The addition of the Fowler flap to a wing appeared to have no appre- 258 Chapter 3: The Design Revolution Document 3-12(a-h) 259

ciable effect on the relation between lift coefficient and angle of downwash. The MODEL calculations in the appendix show that, by proper use of the Fowler flap, the take-off The basic wing was built of laminated mahogany to the Clark Y profile (table I) of an airplane having wing and power loadings in the range normally encountered and had a span of 60 inches and a chord of 10 inches. The trailing edge was cut away in transport airplanes should be considerably improved. and the upper surface replaced by a thin curved metal plate. The lower surface was left open at the rear to serve as a retracting well for the flaps. Blocks were inserted INTRODUCTION to maintain the correct size of well for each size of flap tested. Figure 1 shows the During the past few years the use of flaps on high-performance airplanes as a profile of the wing with the various flaps in place. device for reducing space required in landing has become common. Thus far split The two smaller flaps were made of duralumin to the Clark Y profile and had flaps have been most generally used, probably because of their simplicity of applica- spans of 60 inches and chords of 2 and 3 inches. The largest flap, which is the one tion and their superiority in giving steep gliding approaches and short landing runs: described in reference 2, was made of mahogany and had a 4-inch chord. The flaps the features of flaps with which designers have been most concerned. In order to were supported on the wing by fittings attached to ribs located in the retracting well. retain satisfactory operation from normal flying fields with fast airplanes, however, Several sets of attachment holes in the ribs, combined with several sets of fittings, the use of high-lift devices that improve take-off as well as landing is desirable. Since gave the range of flap positions shown in figure 1. The flaps were supported on the drag is unfavorable to take-off, the comparatively large drag of split flaps places fittings by hinges located at the center of the leading-edge arc of the flaps, angular them among the least promising of high-lift devices in this respect. The Fowler flap adjustment being obtained by set screws attached to the flap moving in quadrantal appears to offer a better compromise between these conflicting requirements. For slots in the fittings. In general, where the term “flap position” is used, the position equal sizes it will give higher maximum lift with no higher profile drag than most of the flap hinge axis is indicated, irrespective of angle, and flap angle is measured other flap arrangements and its comparatively low drag at high lifts is favorable to between the chord lines of the wing and the flap. take-off and steep climb. This effect would normally entail some sacrifice of steep gliding ability. TESTS Although sufficient data to form some estimate of the performance to be expected Five groups of tests were made in obtaining the data presented in this report. from an airplane equipped with Fowler flaps are available (references 1 and 2), they These five groups dealt with maximum lift, optimum flap arrangement for take-off, are inadequate for normal design purposes . The purpose of the tests reported herein standard force tests of optimum arrangements, flap loads, and downwash behind is to provide data to form a rational basis for the design of airplanes equipped with the wing with various flap arrangements. Fowler flaps. It appears that for the Maximum lift.—The maximum lift coefficients obtainable with the 0.20 c and present the purpose will be attained 0.30 c flaps at various positions shown in figure 1 were found by tests in which the by making available the following flap angle was increased from 20° in 10° steps until the peak of the variation of CLmax information: effect of flap size on with flap angle was defined for each position. The range of positions in both cases aerodynamic characteristics attain- was sufficient to surround the point at which the highest lift coefficient was found, able, aerodynamic loads applied to thus isolating an optimum position in each case. Similar surveys had previously the flap in various conditions, and been made with the 0.40 c flap (reference 2) and were not repeated at this time. effect of the flap on downwash. In Optimum take-off arrangement.—Lift and drag data were taken at a range of addition, a convenient method of flap angles between 0° and that giving maximum lift for a series of flap positions estimating the effect of high-lift somewhat more restricted than the range used in the maximum-lift tests. Care was devices on airplane take-off should exercised in these tests also to surround what was judged to be the optimum setting, prove of assistance in cases where both as regards position and angle. this performance feature merits Standard force tests of optimum arrangements.—A series of final force tests, special attention. consisting of lift, drag, and pitching-moment measurements, was made at the flap The tests were made in the positions considered to be of special interest. These included tests of the maximum- 7- by 10-foot wind tunnel of the lift arrangement of each flap, of the optimum take-off arrangement of each flap, and N.A.C.A. (reference 3) at Langley of an arbitrarily selected arrangement representing partial retraction of each flap. Field, Va., during the summer and All tests in these first three groups were conducted in accordance with standard fall of 1934. force-test procedure as described in reference 3. 260 Chapter 3: The Design Revolution Document 3-12(a-h) 261

Flap loads.—Air loads acting on the flaps were found by supporting the flaps foot tunnel, no corrections for these effects have been made. Corrections of several independently of the wing, at the same position and angle as used in the final force sets of airfoil results have indicated that the values of the jet-boundary correction tests of the wing-flap combinations, and by measuring the forces on the wing alone factors, δα = −0.165, and δD = −0.165, used in the standard equations (cf. refer- in the presence of the flap. The flap loads could then be readily computed . In order ence 5) are satisfactory for a 10-inch by 60-inch wing. The static pressure in the jet

to find the center of pressure of the load on the flap, the flap hinge moment was decreases downstream, producing an increment in CD of 0.0015 on normal 12 per- measured by observing the angular deflection of a long slender torque rod required cent c thick rectangular airfoils. Evidence at present available indicates that the effect to balance the flap at the angle in question. Similar measurements of loads and center- of the turbulence in this tunnel is small as compared with the other consistent errors. of-pressure locations on split flaps are more completely described in reference 4. Downwash.—Measurements were made with “pitot-yaw” tubes attached to the RESULTS AND DISCUSSION wing by a rigid support. The reference position thus moved in the air stream as the All test results are given in standard non-dimensional coefficient form. In the angle of attack was changed but remained the same with respect to the wing, as does case of a wing with a retractable surface, the convention of basing coefficients on the the tail of an airplane. The angles of downwash, however, were referred to the initial area that would be exposed in normal flight, that is, the minimum area, has been direction of the free air stream. The apparatus could be adjusted to various horizon- adopted. The coefficients used are then defined as follows: tal distances behind the wing. The pitot-yaw tubes were ordinary round-nosed pitot subscript w refers to the basic wing tubes with two additional nose holes at 45° above and below the tube axis. Alcohol subscript f refers to the flap manometers were used to read the pressures, and the tubes were calibrated in test q = 1/2 ρV2 position in the clear-tunnel air stream. CL = lift/Suq The wind tunnel is of the open jet, closed return type, with a rectangular jet 7 CD = drag/Swq by 10 feet in size. A complete description of the tunnel, balance, and standard force- Cm = pitching moment/cwSwq test procedure appears in reference 3. CNf = normal force on flap (per- Tests were run at a dynamic pressure of 16.37 pounds per square foot, cor- pendicular to flap chord)/Sfq responding to an air speed of 80 miles per hour at standard sea-level conditions. CXf = longitudinal force on flap The Reynolds number {AQ3} of the tests, based on the 10-inch chord of the wing (along flap chord)/Sfq C = flap hinge moment/S c q without flaps, was approximately 609,000. Hf f f ∈, angle of downwash, degrees. PRECISION Maximum-lift condition.—The The accidental errors in the results presented in this report are believed to lie results of the maximum-lift tests are within the limits indicated in the following table: presented as contours showing varia-

tions of CLmax with flap hinge posi- Wing data: Flap load data: Downwash data: tion, irrespective of flap angle. Figures α ±0.10° C ±0.10 ε ±0.5° Nf 2, 3, and 4 show contours for the 20 C ±.05 C ±.06 Lmax Xf percent chord, 30 percent chord, and ± ± Cme/λ .008 CHf .004 40 percent chord flaps, respectively. ± ± ° CD(CL=0) .001 Flap angle .25 Data on the 40 percent chord flap are ± ± CD(CL=1) .004 Flap position .003 taken from reference 2, no further tests ± CD(CL=2) .008 having been considered necessary on Flap angle ±.25° that size of flap after an analysis was Flap position ±.0015c made of the data for the two smaller Consistent differences between results obtained in the 7- by 10-foot wind tunnel flaps. The optimum position is the and in free air may be ascribed to effects of the following factors: Jet boundaries, same for all three flaps: 2.5 percent of static-pressure gradient, turbulence, and scale. In order that the present results be the main wing chord directly below consistent with published results of tests of other high-lift devices in the 7-by 10- the trailing edge. The optimum angle 262 Chapter 3: The Design Revolution Document 3-12(a-h) 263

was 30° for the 20 percent c flap and 40° for the two larger flaps.

Variation of CLmax with flap size is shown in figure 5. The maximum lift coefficient increases approximately in proportion to flap size if the area of only the original wing is considered. This is a reasonably satisfactory basis for comparison of the landing speeds of an airplane with various sizes of flap if a constant maximum speed is maintained. If the maximum lift force that a wing will give at a certain air speed per unit of structural weight is taken as a criterion, it is reasonable to compare the various sizes of flap on the basis of total (wing-and-flap) area. On this basis there is clearly little to be gained by using flaps larger than 30 percent c. Lift, drag, and pitching- moment data for the wing with each of the three flap sizes, with the flap at the setting for maximum lift, are given in figure 6 and in tables III, IV, and V. Coefficients are based on the area and/or chord of the wing alone. The data for the plain wing were obtained with the 20 percent chord flap fully retracted into its well. (See table II.) It is evident that an airplane having a flap of this type would have a much larger range of center-of-pressure travel between various flying conditions than would one with a plain wing. It appears, then, that in a normal type of 2-spar wing the effect of adding a Fowler flap would be to leave the front-spar design load the same as for the wing without a flap but to 264 Chapter 3: The Design Revolution Document 3-12(a-h) 265

increase considerably the design loads on the rear spar. If the speed at which the air- a rigorous selection, comparisons of data that are available (reference 2) indicate the plane may be flown with flap extended is limited to a value reasonably in excess of its choice to be sufficiently near the optimum for practical purposes. landing speed, it appears likely that the loads with flap extended would be reduced Partial retraction of flap.—Lift, drag, and pitching-moment data for the wing to the same magnitude as the largest loads with flap retracted, with flap sizes not with the 20 percent c, 30 percent c, and 40 percent c flaps in a partially retracted in excess of 30 percent c. On this basis it appears that a wing with a Fowler flap as position are shown in figure 17 and in tables VIII to XI. The settings were chosen wide as 30 percent c could be constructed in which there would be no increase in by assuming the flaps to move along an arc from the setting for maximum lift or the weight of the wing structure proper, the only additional weight being due to the optimum take-off to the fully retracted position. The flap hinges crossed the wing flap and its support from the spars. chord line at the 90 percent c station, and the angles at this position were 15° for the Take-off condition.—Investigation of wing-flap combinations to determine 20 percent c flap, 20° for the 30 percent c flap, and 20° and 30° for the 40 percent the flap arrangement most favorable for take-off must involve consideration of per- c flap. Comparison of the characteristics at this setting with those at the maximum- formance parameters of the airplane in question as well as of the aerodynamic effects lift setting shows that the change of characteristics is in the same direction and of of the lifting surfaces. Concurrently with the tests, a series of take-off computations the same order of magnitude as the change of flap setting. was made with the purpose of developing a “take-off criterion” for wings based on Flap loads.—Curves of normal and longitudinal force coefficients, hinge aerodynamic characteristics and depending on airplane design factors to the mini- moments, and center-of-pressure locations of the 20 percent c, 30 percent c, and 40 mum extent possible. The application of such a criterion to the data would then percent c flaps in the maximum lift, optimum takeoff, and partly retracted settings serve to isolate the optimum flap arrangement for take-off. The development of the are shown in figures 18 to 23. The corresponding data appear in tables III to XI. criterion, and associated data, are presented in an appendix to this report. From the magnitude of the load carried by the flap at high lift coefficients of the As the tests and computations progressed, it was found that some general con- combination, it is evident that the flap carries nearly 1½ times its proportionate siderations would serve to isolate the optimum arrangement, without recourse to share of the total load. It appears that this type of flap may be regarded as a separate a rigorous criterion. The computations indicated that normal transport airplanes wing, operating in an air stream whose combined velocity and curvature increase should take off at a lift coefficient greater than 70 percent of the maximum available considerably the load it carries as compared with the load it would experience in the to achieve the shortest run to clear an obstacle . They also indicated that the princi- free air stream. Comparison of load data for a split flap (reference 4) and a Fowler pal aerodynamic characteristics affecting take-off, high lift available, and high L/D flap clearly shows the fundamental difference in the action of the two flaps. At high at the high lift are of nearly equal importance. lifts, the split flap carries almost no lift and offers large drag; whereas the Fowler The wind-tunnel data, plotted as polar curves, are presented in figures 7 to 10 carries a large proportion of the total lift, but with less drag. for the 0.30 c flap and in figures 11 to 15 for the 0.20 c flap. Comparison of these Although this condition is favorable to airplane performance, it implies a large curves on the basis of the considerations previously stated indicated the flap posi- range of center-of-pressure positions for the complete flight range, with consequent tion 0.025 c directly below the trailing edge of the wing, with an angle of 30°, to disadvantages in longitudinal-stability characteristics and possibly also in structure. be the optimum take-off arrangement for both flaps. At this setting each flap has as In connection with structural considerations it is interesting to note that a progres- high ratios of L/D throughout the high-lift region as any other setting tested, within sive reduction in flap loads occurs with increasing flap size if the maximum angle is the limits of accuracy of the tests, and has a higher maximum lift coefficient than kept below 30°. any other setting having as high ratios of L/D. The 40° setting of the 0.30 c flap, at At flap settings giving high maximum lift coefficients, the center of pressure of this same position, gives a higher maximum lift and lower ratio of L/D than the 30° the flap itself has little travel throughout most of the angle-of-attack range and is angle, the percentage difference in L/D being greater than that in maximum lift. generally nearer the leading edge than it would be on an airfoil in a free air stream. Computations (see appendix) verify the conclusion based on the general considera As the flap angle is reduced below 30°, however, the center of pressure moves rapidly tions, that the 30° angle is better with this flap. backward. Lift, drag, and pitching-moment data for the wing with each of three sizes of Downwash.—Some representative data from the downwash measurements are flap, with the flap at the optimum setting for take-off, are given in figure 16 and in shown in figures 24 and 25. Angle of downwash as a function of lift coefficient is tables III, VI, and VII. The choice of the position 0.025 c below the wing trailing shown for two positions behind the wing, with data for the plain wing and for the edge, with a 25° angle, as optimum for the 40 percent c flap is based on the relation same flap settings as were used in the flap load tests plotted on each curve. Only between optimum take-off setting and that for maximum lift of the 20 percent c small consistent deviations from the mean curve, within the limits of test accuracy, and 30 percent c flaps. Although data for the 40 percent c flap are not sufficient for were found for the variety of settings tested. It appears, then, that the addition of a 266 Chapter 3: The Design Revolution Document 3-12(a-h) 267

Fowler flap has no appreciable effect CONCLUSIONS on the basic relation between lift, 1. The maximum lift coefficients, based on area of wing alone, found for the span, and downwash at reasonable three sizes of flap tested were: For the 20 percent c flap, 2.45; for the 30 percent c distances behind the wing. flap, 2.85; and for the 40 percent c flap, 3.17. The maximum lift coefficient for the The foregoing conclusion is sub- wing with flap retracted was 1.31. ject to some question owing to the 2. The location of the flap leading edge for maximum lift was found to be the doubtful nature of the jet-bound- same in all cases, the center of the leading-edge arc being 2.5 percent c directly below ary effect on downwash in the 7- by the trailing edge of the main wing. The flap angles for maximum lift were 30°, 40°, 10-foot tunnel. The corrections in and 40° for the 20 percent c, 30 percent c, and 40 percent c flaps, respectively. this particular case differ consider- 3. The 20 percent c and 30 percent c flaps were found to give the characteristics ably from the theoretical corrections, most favorable to take-off with the same leading-edge location as for maximum lift. probably on account of the combined The optimum angle was 30° in both cases. effect of static-pressure gradient in the 4. The maximum normal-force and longitudinal-force coefficients of the 40 jet and spillage of air over the unflared percent c flap, based on flap area, were 2.89 and −1.25; those for the 30 percent c lip of the exit cone. Different correc- flap were 3.06 and −1.54; and those for the 20 percent c flap were 2.80 and −1.20. tions for different positions of the ref- Center-of-pressure locations corresponding to these coefficients were in each case erence point in the air stream might approximately at the 20 percent c flap chord points. produce greater consistent differences 5. At positions normally occupied by the tail surfaces the relation between lift in downwash between the plain wing coefficient and downwash angle appears from the present tests to be the same for a and flap extended conditions than wing with or without a full-span Fowler flap. are indicated by these tests, though this effect would be small unless the LANGLEY MEMORIAL AERONAUTICAL LABORATORY, variation of the corrections with posi- NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS, tion is greater than seems likely. LANGLEY FIELD, VA., April, 26, 1935. Although the extensive investigation required to establish the cor- Document 3-12(h), Harlan D. Fowler, “Aerodynamic Characteristics of the rections might produce results of Fowler Wing,” Aero Digest 29 (September 1936): 46-50. academic interest, certain effects of combining a variable-lift wing with Aerodynamic Characteristics The Fowler Wing an airplane fuselage would render the HARLAN D. FOWLER results of small technical value. Since a large difference in angle of attack occurs at the same value of CL with different settings of the Fowler flap, a large variation of fuselage attitude and lift at a given Airfoil design has practically reached its maximum, comprehensive and system- wing lift coefficient results from changing flap settings. Thus, at a given over-all atic research by the NACA resulting in such several new highly efficient sections lift coefficient of the airplane, the lift coefficient and downwash of the wing may that possibility of further improvement is remote and of comparatively small value, be expected to change with flap setting. The use of partial-span flaps produces an although costly to discover. effective reduction of span as the flap is extended, causing an additional change of Attempts have been made to so design an airfoil that it would develop especially downwash at constant lift coefficient with changing flap setting. It appears that high L/D ratio, better minimum drag at cruising speed, etc. Earliest efforts were problems involving downwash of variable-lift wings are more susceptible of solution devoted to increase maximum lift (to provide safer landing speed), and to increase by measurement on the actual design in question, rather than by a fundamental drag so as to decrease length of the run when landing. wind-tunnel investigation. Research conducted by NACA for the Bureau of Aeronautics indicated that an airplane equipped with a Fowler variable area wing could reduce landing speed, 268 Chapter 3: The Design Revolution Document 3-12(a-h) 269

decrease landing and take-off runs, and that there would also be indications of Comprehensive research on fowlers of improved climb with the Fowler partially extended. This device was first tried in 20, 30 and 40% chord shows that if the 1927 on a Canuck JN; in 1929 on a Pitcairn; in 1934 at the International Com- maximum angle is kept at or below 30 there petition at Warsaw, Poland, where a similar device was used by Fieseler and Mess- is a progressive reduction in flap loads with erschmitt; in 1935 on a Fairchild F-22 by the U. S. Navy and NACA; and more increasing chord, provided advantage is recently on the Martin Bomber. taken of the corresponding decrease in stall Combined use of variable area, camber, angle of attack and of the boundary speed. Furthermore, if the high speed with layer, requires only an auxiliary airfoil that is extended or retracted as the occasion the fully extended wing is kept below twice demands. The principle is termed a fowler to distinguish it from other flap devices. stalling speed, it is possible to add fowl- Individual aerodynamic properties of each of these principles is not sufficiently great ers having a chord up to 30% to a wing to justify its use alone, as for example, the percent distribution of the maximum lift whose spars have already been designed for coefficient as contributed by the several features given in Table 1. its normal critical loads (except for wings Based on area alone CLmax should vary directly with the fowler chord with it using an airfoil of constant center of pres- fully extended. However, data available show that with a 20% chord fowler set at sure movement) that will be sufficiently -5, fully extended, the increase is slightly higher or about 25% more than for the strong to take the load from the auxiliary normal wing. By virtue of the recess cutout under the trailing edge of the normal wing without appreciably increasing the wing, lift is again increased slightly varying with fowler chord. main wing weight, particularly the rear The major individual increase in maximum lift results from the gap, which spar. Because of the necessity of a recess under the trailing edge of the main wing, appears to increase nearly directly with the chord of the fowler. Variable area is the ribs are consequently shallower and must be strengthened, although not necessarily next most effective factor, with camber a close third. However (by the principle with considerable increase in weight, although there will be a weight increase by involved), the combination of area and camber actually contributes a greater increase adding the fowlers and controls. than the gap, and it is apparent that all three factors have a powerful influence in When the fowler is in its fully extended position, longitudinal force assumes a contributing to the maximum lift, and increase with enlargement of the flap chord. negative value—that is the force tends to close it. At some intermediate point this force becomes zero then tends towards the positive. The control system is therefore designed not to extend the fowlers but to prevent them from closing. With this Partial Span and Lateral Control arrangement and using an automatic hydraulic pressure relief valve system, should Distribution of maximum lift along the wingspan has been investigated and the load become too high, the fowlers will tend to move towards the closed position, checked with actual performance data in cases where the fowler covered a portion of thus preventing undue strain on the main wing structure. However, the hydraulic the wing. The loss of lift is greatest near the fuselage and least at the wing tip. On a device will hold it at any desired position. tapered wing it is hardly worthwhile to extend the fowler beyond 80% of the span Throughout the flying angles of attack, the center of pressure on the fowler as the remaining 20% at the tip represents a loss of about 4% from the full span wing is uniformly forward when in the extended position, at about 30% of its arrangement. However, if the inner end of the auxiliary wing is not extended over chord. Since the main spar of the flap is located at about 20% chord there is no the center 20% span, there will be a loss of about 14% from the maximum. tendency for flutter since the pressure is constantly aft of the point of support. In Considering the small possible gain by extending it to the tip, the question of the case of a twin-engine installation with the propellers directly in front of it, it providing for lateral control means becomes involved. The author maintains that was found that the extended fowler minimized turbulence, due perhaps to the gap due to the powerful lifting capacity of the device, there is no need to sacrifice good action on the boundary layer over the wing. lateral control to attain the last ounce of lift. In this respect, conventional ailerons have been used and the lateral control obtained has been satisfactory, even with the Control Mechanism fowler fully extended. Experiments have revealed that rolling moments are sub- There are two methods of operating the fowlers. The original method consisted stantially higher with it extended than when closed, the ideal condition. In these of fixed tracks, properly curved, attached to the rear spar, with a trolley which is tests, ailerons occupied from 30 to 40% of the semi-span, the smaller span being secured to the auxiliary wing, running along the track. Back and forth motion was adequately effective. accomplished either by a continuous cable or by push-pull rods. The second method 270 Chapter 3: The Design Revolution Document 3-12(a-h) 271

is the rotating or swinging arm which is pivoted to the rear spar with the fowler sus- length of the combined wing and fowler chord, tail surfaces should be at a slightly pended at its outer end. The arm swings back towards the trailing edge, thus extend- greater distance from the center of gravity, or the horizontal tail surfaces made larger ing the fowler and automatically increasing its angle, the latter being accomplished than normal, the area of the elevator being at least equal to one-half the total hori- by a universal joint. Both systems have been tested and each has merits which the zontal tail area. This increase is to be expected because at the very high lift coeffi- other does not possess. The swinging arm presents a perfectly smooth wing when cients obtainable with the fully extended fowler, the downwash angle is considerably the fowler is closed. The fixed track has a small portion of its rear end exposed at all higher. Combined with the low maximum lift coefficient of tail surfaces, this causes times (without causing any appreciable increase in drag) and is less costly to build the tail to stall at certain speeds. For this condition large elevator movements are and lighter in weight than the other method. preferable to stabilizer adjustment, especially on large planes. Also, there is no ten- Speed of extension or retraction is obtained in several ways. Using manual dency to nose over at the take-off or on landing. crank-operated control, movement is fro m 2.5 to 3 inches to one complete turn of Although not completely affirmed, extensive investigations on downwash angles the crank handle. By hydraulic control in one case, movement was accomplished in behind a wing using the auxiliary wing indicate that its addition has no appreciable 15 seconds. There is something to be said about the speed at which fowlers should effect on the basic relation between lift, span and downwash at reasonable distances be operated. They should not be operated too quickly since it gives neither the plane behind the wing. The downwash angle is a direct function of the lift coefficient, but nor the pilot a chance to adjust for the varying conditions. If there is too long an it appears that some correction must be made for the shifting of the angle at zero interval, there will be a decided negative influence on landing and take-off, so that lift as between the normal wing and the extended fowler. In other words, the angle a happy medium on speed of operation is important. of attack for all fowler positions and chord sizes is referred to the basic chord line of Hydraulic and electrical systems are preferable for large planes, the first lend‑ing the normal wing. As it is extended or increased in chord size, the angle at zero lift itself to automatic retraction, if the plane’s speed becomes excessive. Both systems becomes more and more negative. Since the setting of the tail surfaces is referred to are dependable and flexible. the zero lift chord of the main wing, this variation must be considered.

Partially Extended Charac- Landing Distances and teristics Rate of Descent Inherent with the extendable feature The maximum lift coefficient of the device is the transition or partially obtained by a fowler wing is 3.17, extended position. The fowler can be stopped referred to the normal wing area, by at any intermediate position and several such using a 40% chord over the entire span. positions contribute improvement to perfor- By means of an auxiliary nose airfoil this mance. The accompanying diagram illus- can be increased to 3.62. At present, for trates several positions at which it may be set most designs in which the fowlers are to give particularly desirable performances, used, the maximum lift coefficient is the positions shown being spaced in four from 2.00 to 2.50, the lower value being locations, the exact positions for the given attained by using a 20% chord and the type of performance varying with airfoil and latter by using a 30% chord both over 60 airplane design. to 70% of the wingspan. A 40% chord has been used in one wing. The accompanying chart (from NACA Technical Report 534) shows the variation of the maximum lift coefficient with fowlers of various percent chords in terms Longitudinal Stability and Downwash of the basic wing chord. As a high lift device is generally used to increase the maxi- In the normal wing condition, longitudinal stability is as usually obtained on mum speed of a plane keeping a given stall speed, it is proper to consider lift in well-designed planes. terms of normal wing area, with fowler retracted. With the fowler device, as it moves outward and to about 2/3 extended posi- As an illustration of what has been done in full flight tests, some results obtained tion. the change in balance is barely perceptible. In the full out position there is at Langley Field by NACA are given in Table 2, which shows minimum rate of need of more upward elevator movement for landing. Also, due to the increase in descent with the fowler fully extended was about 8.7 ft./ compared to 9 ft./sec. 272 Chapter 3: The Design Revolution Document 3-12(a-h) 273

with the wing closed. Data were obtained on the F-22, results being average of two Seaplane Take-off landings. F-22 has a Fairchild fuselage mounting a specially designed variable area A valuable field of application of the fowler wing is for flying boats, especially wing. The plane was powered by a 95 hp engine. Gross weight was about 1600 lbs., those having long range and large payloads. While the controllable-pitch propel- wingspan 31 ft., normal chord 4 ft. 4 in., normal area 132 ft’, fowler area 27.3 ft’, ler has materially reduced take-off problems for such ships, its combined use with total area 159.3 ft.’ A 30% chord fowler over 71% of the span was used with con- the flap offers further improvements. The problem of a large flying boat is to carry ventional ailerons whose area was about 7.2% of the normal wing area. Wing load a large fuel load plus passengers, mail and equipment, but this usually results in based on normal area was 12.1 lbs./ft 2, and power loading 16.8 lbs./hp. increasing take-off speed. With the higher take-off speed, water resistance of the Running along the ground for take-off or landing, the proximity of the earth hull may become large enough to actually prevent a take-off. With the high lift of causes a reduction in the angle of attack due to induced lift. Also, the higher the lift, the fowler wing, take-off speed is reduced and propeller thrust increased, providing the greater the induced angle of attack of the wing (as this varies directly with the greater excess thrust for acceleration. Even at the full out position, L/D of the wing lift), thus allowing a shorter landing gear. The ground angle being smaller, it is pos- is high enough to cause a substantial reduction in the total water resistance plus air sible to use a true stream‑line fuselage, with a resultant saving in parasite drag. drag. The most effective method of using the fowler is to start extending it about Take-off of Landplanes 10% below take-off speed, so that it becomes fully extended only at the instant of Take-off distances for F-22 are in Table 3 which shows a reduction in total take‑off. Thus the plane starts its run with the normal wing, gets over the hump and distance of 22%. speeds on until at some predetermined speed the fowler begins to extend. Theoretical study of the effect of high lift on take-off, with varying power and Table 4 calculates values for a seaplane of 15,000 lbs. (gross), 1000 ft. normal loading per square foot, showed the fowler wing should be able to reduce total take- area and 1000 hp using a 20% chord fowler occupying 62% of the span. In this case off distance by from 30 to 45%, including the clearance of a 50-ft. obstacle. the time interval between commencing to extend the fowler to the fully extended It should be explained that the fowler is able to contribute to high reduction in position, was about 6 seconds, illustrating the importance of not permitting too take-off run due to the combination of high lift with low drag. The first contrib- long a time for the device to completely open, although a little longer time could be utes to low take-off speed, thus favoring propeller thrust. With low drag, the excess permitted without materially affecting take-off characteristics. thrust is considerably improved. It should also be possible to set the wing in relation to the hull at a lower angle, While the fowler has lower drag and thus less braking effect on landing, the thus reducing hull drag at cruising speed. Another advantage is that the trim angle high lift reduces the landing speed accomplishing results similar to the split flap. of the hull can be maintained nearly constant at get-away because the fowler will However, it should be pointed out that due to the high LID ratio with the fowler pull it off. wing it is necessary for the pilot to use only ordinary flight technique. This is par- The fowler wing can also be utilized to increase fuel load with the same wing ticularly desirable in forced landings. area, or alternatively, the same fuel capacity can be kept and the wing area reduced. Climb and Ceiling Wind tunnel research on a complete model indicates that with the fowler extended about 25%, L/D ratio is higher and occurs at a larger lift coefficient. When applied to a given design, service ceiling can be improved. With planes supercharged to high altitudes, the lift coefficient is larger, and with smaller drag the ceiling is bettered. This characteristic will not be detectable near sea level, as compared with the normal wing, but it should appear at altitudes. Flight tests on F-22 at 2000 ft. indicated that, with the fowler extended 1/3, rate of climb was as good as with it closed, but the angle of climb was higher. On another plane it was found that at 9000 ft. rate of climb was appreciably higher than for the normal wing, and at low speed the gain was about 10%. Also, test results seem to indicate the possibility that when one of two engines fails that, by partially extending the fowler, both rate of climb and ceiling can be improved.

Document 3-13 275

Document 3-13

Standard Steel Propeller Company to Ryan Airlines, undated (ca. April 1927), in Charles A. Lindbergh, The Spirit of St. Louis (New York: Scribner, 1953; reprint, St. Paul: Minnesota Historical Society Press, 1993), p. 116.

As with all other aspects of the design of the Spirit of St. Louis, Charles A. Lind- bergh carefully calculated the role of his airplane’s propeller before setting off on his historic transatlantic flight from New York to Paris. In the spring of 1927, the Standard Steel Propeller Company of Pittsburgh produced four nine-foot-diameter dural (aluminum alloy) ground-adjustable propellers for Lindbergh’s Ryan-built airplane; the company built them according to “specification No. 1519” and sent them on to Ryan Airlines in San Diego. Inspected by Standard Steel’s assembly fore- man and inspector, Alexander F. Manella, the propellers consisted of twelve pieces designed for use with the Wright Whirlwind J-5C radial engine. Lindbergh and Donald A. Hall, the chief engineer for Ryan Airlines, asked Standard Steel to recom- mend a specific blade pitch that would help them achieve the range needed to reach Paris. The word received back came in the telegram reproduced below. What Standard Steel did was simply calculate which propeller blade angle was best, for the propeller itself was essentially “off the shelf.” Acting on Standard Steel’s recommendation and influenced by urgent time constraints, Lindbergh accepted the compromise setting that favored a setting optimum for cruising conditions. This was a big gamble because it was a pitch setting for the propeller that provided minimum efficiency for takeoff. Still, the metal ground-adjustable propeller offered him a flexibility that wooden and metal fixed-pitch propellers could not. With its propeller blade set at 16.25˚, the Wright J-5C engine produced 190 hp at 1,545 revolutions per minute; this generated a static thrust of 700 lbs. The maximum airscrew efficiency of the Spirit’s propeller rated out at 74 percent. Roosevelt Field on Long Island, from which Lindbergh lifted off for Paris, was over a mile long and the only suitable place for heavy takeoffs in the New York area. But obstacles surrounded the field, telephone wires and low tree-covered hills, dangers of which Lindbergh was well aware. As he expected, the Spirit of St. Louis required almost every inch of the field to lift its 5,000-lb load (mostly fuel) into the air. Years later, Lindbergh recalled his approach to the problem: “I decided to ‘feel’ the plane off the ground just as I had often done in underpowered planes while barnstorming. If I felt I was not going to get off the ground, my plan was to simply cut the throttle” (quoted in Cassagneres, Spirit of Ryan, p. 54). Perhaps more than any other element of risk in an otherwise carefully planned flight, Lindbergh’s take- off threatened failure, even disaster. Fortunately, as it happened, the straining Spirit 276 Chapter 3: The Design Revolution lifted off the ground about 1,000 yards in front of the telephone wires at the end of the field—although the angle of the newsreel pictures made it look a lot closer. Thus, one might call Lindbergh’s takeoff the most dramatic and the most decisive event of his entire flight, for as he wrote in his book The Spirit of St. Louis (Scribner’s Sons, 1953): “My propeller is set for cruising, not for takeoff” (p. 183). This passage perhaps illustrates other innovations as well, especially flaps.

Document 3-13, Standard Steel Propeller Company to Ryan Airlines, undated (ca. April 1927), in Charles A. Lindbergh, The Spirit of St. Louis (New York: Scribner, 1953; reprint, St. Paul: Minnesota Historical Society Press, 1993), p. 116.

WESTERN UNION

HOLMSTEAD [SIC] PENN.

RYAN AIRLINES SAN DIEGO, CALIF.

15.5 DEGREES SETTING PROBABLY NECESSARY ON YOUR MONO- PLANE TO GET TAKEOFF WITH HEAVY LOAD FUEL ECONOMY WILL BE IMPROVED ON HIGHER PITCH SETTING STOP IF TAKEOFF IS SATISFACTORY WITH 15.5 SETTING SUGGEST TRY 16.5 AS THIS WILL IMPROVE FUEL ECONOMY.

STANDARD STEEL PROP. CO. Document 3-14 277

Document 3-14

C. B. Allen, “Hamilton Standard Wins Collier Trophy for Controllable Pitch Propeller,” The Bee-Hive 8 (June 1934): 1-2.

In 1933, the National Aeronautic Association awarded its Collier for the year’s greatest achievement in American aviation to Hamilton Standard Propeller Com- pany, with particular credit to Frank W. Caldwell, chief engineer, for development of a controllable-pitch propeller that was then entering general use. Caldwell’s design controlled its blades via bevel gears that were turned by a drumcam in the propeller hub. Hamilton Standard later went on to refine the design by adding a “constant speed unit” or CSU. This gadget controlled the propeller automatically to permit the desired engine speed. The aircraft industry continued to make improvements. By late in World War II, the controllable-pitch propeller added reversible-pitch settings that further slowed an aircraft for landing. Even as the turbojet revolution gained momentum, aeronautical engineers continued to perfect propeller applications for reciprocating engines. Airplanes such as the Boeing B-29, Boeing B-50, Lockheed Constellation, and Lockheed C-130 Hercules all employed large and sophisticated controllable-pitch propellers of three and four blade sections. Without advanced forms of this type of propeller, piston-engine airliners such as the Douglas DC-6 and DC-7 of the late 1950s and early 1960s could not have achieved effective performance. Without question, the variable-pitch and controllable-pitch propellers made essential contributions to the reinvention of the airplane inherent to the airplane design revolution of the interwar period.

Document 3-14, C. B. Allen, “Hamilton Standard Wins Collier Trophy for Controllable Pitch Propeller,” The Bee-Hive 8 (June 1934): 1-2.

PRESENTATION last month by President Franklin D. Roosevelt to the Ham- ilton Standard Propeller Company, of East Hartford. Conn., through its chief engineer, Frank Walker Caldwell, of American aviation’s most coveted annual award-the Collier Trophy for 1933-has met with wide approval from the industry because of the practical benefits it has enjoyed from the first practical “gearshift of the air.” More than 500 controllable pitch propellers of the type for whose development the award was made are now in everyday service, and their users will have no complaint with the National Aeronautic Association for selecting this device as “the greatest achievement in aviation in America, the value of which has been demonstrated by actual use during the preceding year.’’ 278 Chapter 3: The Design Revolution Document 3-14 279

Eliminates Faults of Old Models There have been three eras in the development of the airplane propellers—the Essentially the controllable pitch propeller supplies an airplane with a “low age of wood, which lasted through the World War and seven years beyond, the metal gear” for takeoff and climb purposes and a “high gear” for economical cruising. It propeller (both with fixed blades and blades that could be adjusted to any desired eliminates the chief faults of the fixed pitch propeller--the fact that its blades “bite” angle before flight, but not altered in the air), which held sway some nine years and so much air as the plane moves slowly over the ground the engine cannot “turn up” won the Collier Trophy for Dr. S. Albert Reed in 1925, and now the “gearshift of its full rated horsepower or “race” impotently once the ship is in level flight, if the the air” that may be changed at the pilot’s will in mid-air. blades are set at a flat angle designed to give a reasonably quick take-off, because The 1933 committee of award, appointed by President Hiram Bingham of the they do not get a sufficient grip on the medium through which the whole ship is National Aeronautic Association, consisted of Rear Admiral Emory S. Land, U.S. moving at so swift a pace. Navy: Colonel Edgar S. Gorrell, Air Corps Reserve, and Chief of Staff of the Air Obviously, such a development always has been highly desirable in aviation, Corps in France during the war: Colonel Clarence M. Young, former Assistant Sec- but without it the high-speed transport planes that have made their appearance on retary of Commerce for Aeronautics: General Frank Hitchcock, who, as Postmaster America’s air lines during the last year would not have been possible. No fixed pitch General, arranged for the first flight of air mail in America in 1909, and Earl N. propeller could be expected to function with any sort of efficiency over the wide Finley, Editor of U.S. Air Services. range that lies between the take-off and top speeds of these craft, and such devices as the wing-flap “air brake” which enable 200-mile-an-hour transports to land and Notables at Presentation take off at fifty miles an hour would be pretty much in vain save for the controllable Among the guests who witnessed the presentation were two previous holders: pitch propeller. Harold Piteairn of Philadelphia and Glenn L. Martin of Baltimore. Others present included President Hiram Bingham of the National Aeronautic Association; Eugene “Gearshift of the Air” L. Vidal, Director of Aeronautics of the Department of Commerce; J. Carroll Cone, Also, the safety features of the multi-motored airplane, especially the two-engine Assistant Director of Aeronautics; Dr. George W. Lewis, Director of Research of variety, would be greatly impaired without the “gearshift of the air” for the reason the National Advisory Committee for Aeronautics; Don L. Brown, President of that, when one power plant fails, the plane inevitably slows down in the air and only the Pratt & Whitney Aircraft Co.; Ray Cooper, General Manager of the National by putting the remaining sound engine in “low gear,” where it can “turn up” its full Aeronautic Association; William R. Enyart, Secretary of the Contest Committee of rated horsepower (and in some cases deliver for short periods more power than its the N. A. A.; Joseph Edgerton and Bob Ball, newspaper men. normal rating), is it possible to continue flight and maintain altitude--particularly when the failure occurs just after leaving the ground or during flight over high-altitude terrain. United Air Lines alone has more than 100 Hamilton Standard controllable pitch, propellers in service on the Boeing monoplanes it is operating over its transcontinental and other routes. Some of these have flown 1,500 hours and more--the equivalent of 225,000 miles flying-without failure of any sort. The device is simple and rugged, a piston, operated by oil pressure from the engine, twisting the blades into “low gear” when the pilot pulls a lever in the cockpit. Once he is in the air and has gained the level at which he wishes to fly, he releases the lever and the blades are automatically pulled into “high” by two counter-weights, which the powerful, outward-pushing oil piston heretofore has thrust and held back.

Years of Development Work Mr. Caldwell designed and patented the controllable pitch propeller in 1928, but years of development work and $200,000 in cash were required before all of its mechanical difficulties had been conquered and its creator was satisfied to put it into production a year ago last January.

Document 3-15(a-e) 281

Document 3-15(a-e)

(a) Smith J. DeFrance, “The Aerodynamic Effect of a Retractable Landing Gear,” NACA Technical Note 456 (Washington, 1933).

(b) Hal L. Hibbard, “Problems in Fast Air-Transport Design,” Mechanical Engineering 55 (October 1933): 611-617.

(c) “Preliminary Study of Retractable Landing Gears for High and Low Wing Monoplanes,” Air Corps Information Circular 7 (18 February 1933): 1-9.

(d) Richard M. Mock, “Retractable Landing Gears,” Aviation 32 (February 1933): 33-37.

(e) Roy G. Miller, letter to editor, Aviation, “Retractable Landing Gears,” Aviation 32 (April 1933), with response from Richard M. Mock: 130-131.

One of the most significant “shelf items” required to reinvent the airplane was retractable landing gear; as essential as it was, it also proved one of the most evasive. The earliest known concept for a retracting mechanism dates from November 1911 and to an obscure aeronautical inventor by the name of F. McCarroll who received a U.S. patent for his invention in November 1915. The first airplane to appear with such an undercarriage was the Dayton-Wright RB-1 of 1920, which also possessed the world’s first variable-camber wing. In 1922, the idea of retracting gear got a major shot in the arm as part of Louis Breguet’s agenda for aircraft streamlining (see Document 3-4). As discussed in the text, the Verville-Sperry R-3 racer of 1923 brought a lot of attention to retractable landing gear, which it incorporated, when it flew in the Pulitzer Trophy Race of 1923 in St. Louis, winning the race in Dayton the following year. Systematic testing of the concept did not really begin until 1927, when research engineers at the NACA’s laboratory at Langley Field began looking into the relation- 282 Chapter 3: The Design Revolution Document 3-15(a-e) 283 ship between fixed landing gear and aerodynamic drag. The NACA tests, conducted tests,” which backed up his conclusion. What he was no doubt referring to was a test in Langley’s PRT on an army-owned Sperry Messenger airplane with its outer wing program conducted at the GALCIT wind tunnel at Caltech on a Northrop Alpha panels removed, also represented the world’s first wind tunnel tests of a full-scale airplane employing streamline coverings known as “spats” or “pants.” These were aircraft, not just a scale model or individual components. These tests in the PRT teardrop-shaped fairings around the wheels of a fixed landing gear and enclosing the revealed that the resistance of the protruding fixed landing gear amounted to an landing strut and wheels, which helped reduce drag. Spats had become popular with astounding 40 percent of the airplane’s total drag. The tests defined the “real magni- the Lockheed Vega and most fixed landing gear following the Vega used them. tude of the landing gear drag penalty” and provided precisely the kind of empirical Despite the dramatic benefits that retractable landing gear offered, widespread data needed to encourage the American aviation industry to devote more resources development and adoption was slow. As late as 1933, as we see in the third docu- to the design of landing gear. This proved to be a critical step in the process of rein- ment in this string, the U.S. Army Air Corps had no clear idea which forms of venting the airplane, with the NACA becoming more and more involved in evaluat- retractable landing gear were feasible. The authors of the Air Corps information ing the value of retractable landing gear for industry. circular continued to stress the advantages of fixed gear with streamlined wheel The following string of five documents demonstrates the growing commitment fairings even for high-performance aircraft over those possible with retractable gear. to retractable gear. The first concerns NACA tests of the Lockheed Altair mono- The fourth article below is a February 1933 trade journal article that discussed the plane in Langley’s Full-Scale Tunnel for the purpose of learning the advantages and myriad of variables related to retractable landing gear design. Its author reported on disadvantages of the new type of gear and specifically whether or not the gear con- industry’s attitudes toward the mechanical difficulties it was facing when trying to tributed to the aircraft’s excessive takeoff run. NACA researcher Smith J. DeFrance integrate retractable gear into aircraft design. The string closes with a letter to the (future director of the NACA’s Ames Aeronautical Laboratory in California, which editor endorsing the tangible benefits that retractable landing gear would provide came to life in 1941) found that the retractable landing gear improved performance to aircraft operators. generally in flight but had negligible effects on takeoff and landing. DeFrance and The superiority of retractable landing gear grew more and more obvious as other NACA engineers understood that, before the industry could wholeheartedly synergistic advances related to power plants, aerodynamics, structures, and stability accept retractable landing gear, precise knowledge of the flow of air caused by land- and control came together through the 1930s, boosting the speed and performance ing gear when it was extended had to be in hand. When the gear was retracted, the of aircraft to unprecedented levels. By the start of World War II, the advantages of aerodynamics were better, but when the gear was extended, the aerodynamics then retractable landing gear were clear, and the U.S. aircraft industry had turned retract- became problematic. Until the gear performed more effectively overall, it was dif- able landing gear into its standard. ficult to overcome the inertia favoring the certainties of traditional fixed gear, such as that still being put even on innovative aircraft like the Northrop Alpha. Document 3-15(a), Smith J. DeFrance, “The Aerodynamic Effect of a The second document in the string reflects the type of thinking that opposed Retractable Landing Gear,” NACA Technical Note 456 (Washington, 1933). the use of retractable landing gear. Similar to the objections to using the NACA cowling, many individuals in industry did not feel that retractable landing gear was THE AERODYNAMIC EFFECT OF A RETRACTABLE LANDING GEAR the most practical solution. One industry representative who thought this way was By Smith J. DeFrance Hal L. Hibbard, assistant chief engineer of the Lockheed Aircraft Corporation. In 1933, Hibbard presented the following article at the joint meeting of the Society of Automotive Engineers and the American Society of Mechanical Engineers held SUMMARY in conjunction with the 1933 National Air Races in Los Angeles in July. Examin- Tests were conducted in the N.A.C.A. full-scale wind tunnel at the request of ing all the components that made up a complete airplane, Hibbard addressed the the Army Air Corps to determine the effect of retractable landing-gear openings aerodynamic differences between fixed and retractable landing gear, differences that in the bottom surface of a wing upon the take-off characteristics of a Lockheed he considered to be “greatly exaggerated.” The major source of drag in landing gear, Altair airplane. The tests were extended to include the determination of the lift and according to Hibbard, was the aerodynamic interference caused by parts of the land- drag characteristics throughout the angle-of-attack range with the landing gear both ing gear and its relationship with the wing and fuselage. If designers would only retracted and extended. carefully streamline their fixed landing gear, then the difference between fixed and Covering the wheel openings in the wing with sheet metal when the wheels retractable gears would be negligible; in fact, the overall advantage could even go to were extended reduced the drag only 2 percent at a lift coefficient of 1.0, which was fixed gear. At one point in his presentation, Hibbard referred to “careful wind tunnel assumed for the take-off condition. Therefore, the wheel openings in the bottom 284 Chapter 3: The Design Revolution Document 3-15(a-e) 285 side of the wing have a negligible effect upon the take-off of the airplane. Retracting effect upon the lift and drag characteristics when the gear is extended. the landing gear reduced the minimum drag of the complete airplane 50 percent. Langley Memorial Aeronautical Laboratory, National Advisory Committee for Aeronautics, Langley Field, Va., March 16, 1933. INTRODUCTION A somewhat excessive length of run had been required by a certain low-wing REFERENCE monoplane, a Lockheed Altair, during take-off. This airplane has a landing gear that 1. DeFrance, Smith J.: The N.A.C.A. Full-Scale Wind Tunnel. T.R. No. 459, is completely housed in the wing when retracted; and when it is extended, open- N.A.C.A., 1933. ings having an area equal to the side area of the struts and wheels are exposed on the lower surface of the wing. It was desired to know whether these openings were Document 3-15(b), Hal L. Hibbard, “Problems in Fast Air-Transport Design,” causing the detrimental effect upon the take-off characteristics. Consequently, at Mechanical Engineering 55 (October 1933): 611-617. the request of the Army Air Corps tests were conducted upon this airplane in the N.A.C.A. full-scale wind tunnel to determine the effect of the wing openings upon the lift and drag characteristics. Problems in FAST AIR-TRANSPORT DESIGN TESTS AND APPARATUS by Hall L. Hibbard The airplane was mounted in the wind tunnel as shown in Figure 1. The lift and drag forces were measured with the wheels retracted, with the wheels extended and In 1910 the average speed of the fastest railroad trains in this country was 34.6 wheel wells open, and with the wheels extended and the wells covered with sheet- miles per hour. In 1920 this average dropped to 34 and in 1930 it was 40.9. Thus, metal. The tests were conducted at an air speed of approximately 60 miles per hour. over a period of twenty years, the average speed of the nation’s thirty fastest trains Figure 2 shows the landing gear extended and the wheel wells open. A description increased 6.3 miles per hour. This rather unimpressive change in speed has been of the tunnel and balance equipment is given in reference 1. due in a large measure to the continuously increasing traffic congestion faced by all surface transportation. Increased speed has been fraught with danger. And here is DISCUSSION OF RESULTS one of the many advantages that air transport has over all others. Increased speeds, The primary purpose of the investigation was to determine the effect of the almost without limit, are possible with no attendant dangers and even added safety. landing-gear openings in the wing upon the lift and drag characteristics during take- Air transport, even during its short existence, has shown increased speeds of 30 to off. Assuming an angle of attack for take-off that would give a lift coefficient of 1.0, 50 miles per hour, and it has barely started. and comparing the curves on Figure 3 for conditions with landing gear extended It would be impossible in the brief space allotted to this paper to discuss all and wheel wells both open and closed, it can be seen that closing the wells reduced points of transport design. Therefore, a few, dealing particularly with increasing the the drag coefficient only 2 percent. Therefore it can be said that the openings in the speed of transport aircraft, have been chosen. wing for the retractable landing gear of the Lockheed Altair have a negligible effect In connection with the discussion that follows, it will be of interest to have in upon the take-off of the airplane. mind a hypothetical airplane, and to note what effect certain changes have on it. Let Because of its retractable feature, the landing gear on the Lockheed Altair is us assume, therefore, that this hypothetical airplane has a gross weight of 7500 lb, a not aerodynamically clean. It is interesting to note, however, that retracting the horsepower of 700, a wing area of 400 sq ft, a span of 50 ft, giving an aspect ratio of gear reduces the minimum drag of the complete airplane 50 percent. It is therefore 6.25, a wing loading of 18.75 lb per sq ft, and a power loading of 10.70 lb per hp. apparent that, if the landing gear proper is streamlined to reduce the drag, the take- Let us further assume that this is a low wing cantilever monoplane with retractable off characteristics will be improved; but such a change, if carried to the extreme, gear, controllable pith propeller, but with no wing flaps. would have a detrimental effect upon the landing characteristics and might necessitate the installation of some device to reduce the landing distance required. THE PROPELLER CONCLUSIONS Before high-speed airplanes became so numerous and slow-speed planes were A retractable landing gear of the type on the Lockheed Altair may, when in vogue, aircraft propellers were fairly satisfactory.The speeds were not so high but extended, account for 50 percent of the minimum drag of an airplane, but the that added safety. Air transport, even during its short existence, has shown increased openings that house the gear in the bottom surface of the wing have a negligible speeds of 30 to 50 miles per hour, and it has barely started. 286 Chapter 3: The Design Revolution Document 3-15(a-e) 287

It would be impossible in the brief space allotted to this paper to discuss all that the increase in speed was not what it should have been, because of the increase points of transport design. Therefore, a few, dealing particularly with increasing the in frontal area. speed of transport aircraft, have been chosen. In single-motor planes, there is but one location for the power plant, namely, In connection with the discussion that follows, it will be of interest to have in the nose of the fuselage. In multi-motor designs, if the engines are to be placed in mind a hypothetical airplane, and to note what effect certain changes have on in the wings, the motors should be placed so that the propeller is located directly it. Let us assume, therefore, that this hypothetical airplane has a propeller set at ahead of the leading edge of the wing by an amount equal to 2.5 percent of the a pitch to give best performance at high speed also had satisfactory take-off and chord. In this connection it is of interest to note the tests which the N.A.C.A. is at climb characteristics. On high-speed planes, however, a propeller with a fixed pitch present performing in connection with air-cooled versus liquid-cooled engines with will not have satisfactory characteristics for both take-off and high speed. One or radiators in nacelles. The results indicate a drag of 74 lb with the air-cooled arrange- the other must be sacrificed. The problem has become so acute that in some cases ment and drags of 72, and 76 lb, respectively, with two liquid-cooled installations. where speed could not be sacrificed the Department of Commerce requirements for This translated into cruising speed of one of the later type bimotor transports would take-off have limited the gross weight of the airplane. Such was the case two years mean the same cruising speed for either installation. This was also verified by the ago with the first Lockheed Origin which was equipped with an engine without a Army some months ago in the case of two identical bimotor low-wing bombers, supercharger. The take-off was so poor that even though the airplane was capable one with air-cooled and the other with liquid-cooled engines. The speeds of the two structurally of carrying a higher gross weight, this was not allowed. Of course, the were almost identical. propeller pitch could have been changed to secure a satisfactory take-off, but tests Power plants in air transports should be supercharged. Transport airplanes showed that the cruising speed was so reduced of the planes entered in the present rarely cruise at sea level. Supercharging will give the rated power at altitude and air races. The engine diameter was greater than that of the fuselage, and the cowl high output in emergencies at sea level for short periods of time. was carried straight back. Another cowl was built which was drawn in at the rear. Speed tests were run with both types of cowl and the plane was approximately 15 LANDING GEAR miles per hour faster, when fitted with the cowl drawn in at the rear. In the case of a The gain in the speed of an airplane with a retractable landing gear over one plane with a larger fuselage, cowls extending straight back were found to be best. equipped with a well-streamlined gear of the non-retractable type has been greatly The first tests completed by the N.A.C.A. indicated that a double nose section exaggerated. A great increase in speed because of retraction of the gear usually means in the cowl was beneficial. Later tests, however, showed that increased speeds could that the original landing gear design was faulty. Furthermore, after retractable gears be obtained by omitting this inner section of the nose cowl and changing the shape are installed, there is always the tendency to compare speeds with the gear up and slightly. British tests and some private tests in this country have verified this conclu- down, which is obviously unfair, as retracted landing gears are never faired or stream- sion, but in spite of this fact, some manufacturers still continue to use the double lined and arc usually quite unsightly. section which not only reduces the speed but is also more expensive. Take, for example, the Lockheed Orion. This is the original Sirius model For speed and cooling, it is not the quantity of air passing through the cowl that equipped with a retracting gear. The major cause of drag in landing gears is interfer- is important but the velocity and distribution. It is a common error to assume that ence. In the Sirius landing gear, all struts arc nicely faired, fairings are placed over blocking off the air at the nose decreases the drag. Such is not the case. Air must pass the wheels, and all joints to fuselage and wheels are well filleted. But the parts are through the cowl in order to give the least drag. Tests show that the drag is higher too close together, angles are too acute, and the gear is close to the underside of the when the nose shutters arc closed than when they arc open. Much is to be learned wing.The result is excessively high interference. Thus, with the retractable gear, the about the proper distribution of air under N.A.C.A. cowls, and further tests should increase in speed was more than 2.5 miles per hour. be conducted along this line. Careful wind-tunnel tests show conclusively that gears in which the interference has been removed add but little to the drag of the airplane. This type of gear is a POWER PLANTS pure cantilever type without external bracing of any kind, similar to that pioneered From the point of view of the airplane designer, the engine diameter should by Northrup. Care must be taken in the design of this gear to use a good streamline be as small as possible. In this connection it is gratifying to note the progress that shape and it is important that the gear be carefully faired at the top into the wing is being made along this line with all types of air-cooled engines, including the or fuselage. A cantilever gear of this type will reduce the high speed of the airplane two-row radial type. Aircraft manufacturers have had the experience of substituting less than 6 miles per hour, which means a reduction in cruising speed of 3 to 4 miles engines of greater horsepower and larger diameter, only to find in almost every case per hour. 288 Chapter 3: The Design Revolution Document 3-15(a-e) 289

Thus it can be seen that there is some question as to the advisability of the Document 3-15(c), “Preliminary Study of Retractable Landing Gears for High retractable gear from the standpoint of speed increase. However, the retractable gear and Low Wing Monoplanes,” Air Corps Information Circular 7 (18 February is still recommended as manufacturers will not install the non-interference type 1933): 1-9. gear. They will read this article, or others, be much pleased and relieved to know that gear retraction gives such small increase in speed, and then install on the very PRELIMINARY STUDY OF RETRACTABLE LANDING GEARS FOR next design the old type of landing gear with all its speed-reducing interferences. HIGH AND LOW WING MONOPLANES When retractable gears first came on the market, rumors were spread about the (Prepared by E. H. Schwartz, Materiel Division, Air Corps, Wright Field, Dayton, terrible results that would surely ensue if an airplane was forced to land with the Ohio, October 29, 1932) gear retracted. Up to the present, six cases arc known where Lockheed plants have been forced to land with the gear retracted. In no case have there been any injuries SUMMARY to occupants, and in no case have more than minor damages resulted from the landing. The airplane usually slides 150 to 250 ft, depending on the condition of This report sets forth the results of an investigation made to determine whether the ground. Because of the very high cushioning effect when the wing is only a foot or not the use of the retractable landing gear is advantageous in view of the large or so from the ground, the landing is not rough or severe.In other words, instead number of difficulties encountered in its design, operation, and maintenance. The of being dangerous, retractable gears have become a safety feature and transport investigation also included cantilever landing gears with streamlined wheel fairing operators arc demanding them from that standpoint. It is easy to see that safer ice and fixed streamline landing gears. The results of this study indicate that the retract- landings can be made in snow, in marsh lands, and on water; and in mountainous able landing gear can be used advantageously on the multi-engined monoplane with country landings can be made on very small patches of ground. the landing gear retracting into the engine nacelle or wing, and that the cantilever landing gear with wheel fairing is best suited for pursuit, attack, 2-place observation, or single-engine cargo airplanes. These landing gears permit use of monocoque FUSELAGE AND ACCESSORIES fuselage structure. Fuselages for fast airplanes should be as small as possible, consistent with com- A number of retractable, cantilever, and fixed streamline landing gears were fort. There is need for wind tunnel data on fuselage sizes as there is little informa- examined. The indicated increase in high speed due to retraction of retractable land- tion on this subject. The British in the Schneider Trophy racers found that decrease ing gears is misleading, as in all cases it was found that the retractable landing gear in frontal area did more to increase the speed than any other item. On the Super was not streamlined, and large wing or fuselage openings prevailed with the landing marine racer, the cross-section area was reduced from 8.46 to 8.I8 sq ft, with a gear in the down position, giving a low high speed with wheels down. This accounts resulting increase in speed of 11 miles per hour. This was in spite of the fact that the for the comparatively smaller speed increase obtained from wind tunnel tests on resistance per square foot of cross-sectional area was higher for the plane with the removal of faired landing gears. A study was also made of the weights of the various smaller cross-sectional area. In a comparison of two of last year’s racers, it was shown types of landing gears on present United States Army airplanes for comparison. that the drag at 100 miles per hour for the one with the larger fuselage was 89.7 lb, The number of retractable landing gears of the fuselage and wing type procured while that for the other was 80.4 lb, notwithstanding the fact that the latter had 30 by the Air Corps is small. For this reason, little work has been done up to the pres- square feet additional wing area.The straight away speeds of the two airplanes with ent on improving the design of retractable landing gears. Therefore, a study was the same horsepower were 2.47.3 and 2.77 miles per hour, respectively. It would made of possible retractable landing gears, and, as a result, the landing gears shown appear then that there is a limit to fuselage size, and from most indications, the in Figures 1 to 10, inclusive, were conceived, and it is believed that they have good smaller the faster. possibilities for future designs. Another factor which should be taken into account in fuselage design and which This study also showed that a good installation of a retractable landing gear to has not been given enough consideration is the shape of the fuselage across the wing. an existing airplane cannot be made efficiently or economically without a complete Any element, even a flat sheet, placed vertically and parallel to the airstream will redesign of fuselage or wing. interfere with the airflow on the upper surface of a wing. There is a definite flow of All weight data on landing gears was obtained from the actual weight state- air over the upper surface of cantilever wings, such as are being used on present-day ments AN-9102 and AN-9103 furnished the division by the contractors. All air transports, which should not be disturbed. It is evident that if a flat sheet affects the plane weights were taken from the 689 inspection reports of the various airplanes. airflow, the modern fuselage must make a great disturbance. The fuselage should All data on the high speed of airplanes and models was obtained from official conform in some degree to the direction of airflow set up by the wing. flight and wind-tunnel tests with the exception of the YlC-17 airplane. 290 Chapter 3: The Design Revolution Document 3-15(a-e) 291 292 Chapter 3: The Design Revolution Document 3-15(a-e) 293

OBJECT The object of this report was to determine whether or not the use of the retractable landing gear is advantageous in view of the large number of difficulties involved in its design, operation, and maintenance, and to determine the type of landing gear which will give the best installation and performance on various types of monoplanes in use by the Air Corps.

DISCUSSION In this investigation the YlC-12 airplane has been considered as a typical high- wing monoplane for the possible installation of a retractable landing gear. Similarly the YlC-23 or XP-900 may be considered in the same class for low-wing monoplanes. (Gross weight 4,000-5,000 pounds.) XO-35 or XB-9 considered typical of engine nacelle type landing gear. It must be realized that the application of a retractable landing gear to an existing airplane will involve considerations and result in decisions that would differ somewhat from those considerations involved at the time of layout of a new design, for in the latter case fuselage shape, bulkhead locations, spar locations, wing thickness, etc., can be readily laid out or altered to best suit a retracting mechanism. At the start of the investigation certain governing factors must be recognized, namely (1) That the landing angle of the airplane be as large as possible, of a magnitude to enable the airplane to develop not less than 90 percent of its maximum lift coefficient in landing. A decrease in landing angle will result in an increased landing speed. (2) That the angle in side view between a line passing through the center of gravity of the airplane and the center of the wheels, with the airplane in the level landing attitude, and the vertical must not be less than 12° for airplanes without brakes and must be 20° for airplanes with brakes. (3) That the angle in the front view between the vertical and a line joining the center of gravity and the point of contact of either wheel and the ground shall not be less than 25°. (4) That the propeller clear the ground by at least nine (9) inches with the airplane in either a flying or landing position. The foregoing requirements are taken from the Handbook of Instructions for Airplane Designers, Volume I. The purpose for which the airplane is to be employed may establish other requirements, such as: (1) That the landing gear in its retracted position offer no interference with the pilot’s location. (2) That the cargo or bomb space remain intact in cargo or bomber airplanes. (3) That certain space in the vicinity of the center of gravity, or elsewhere, be reserved for military equipment, etc. 294 Chapter 3: The Design Revolution Document 3-15(a-e) 295

(4) That the fuselage section be as small as possible to give the best vision pos- length or travel of the landing gear. The landing gear length must be such so as to sible. meet the requirements of the Handbook as stated on page 1. This type landing There are other general requirements to be met by retractable landing gears, gear is unsuitable for pursuit, attack, or 2-place observation airplanes, since fuselage again possibly subject to alteration in the case of special-purpose airplanes: space is very limited and since a small fuselage section is most important for vision (1) That the handling of the airplane on the ground be as simple as possible and performance. Due to the large cut-outs on the sides and bottom of a critical sec- (compared with a 2-wheel landing gear with swiveled tail wheel). tion of the fuselage, an inefficient structure must be used to carry the required loads. (2) That the retracting gear and movement of parts be such as to permit a struc- An inefficient structure means an excess structure weight. The large cut-outs neces- ture capable of carrying the fuselage design loads, and that this structure preferably sary for retraction of this landing gear prohibits the use of a monocoque structure. be determinate and simple. The wing type of retractable landing gear is best suited for a low-wing mono- (3) That the retracting mechanism be reliable and incapable of jamming in any plane, which requires only a short landing gear, since the length of landing gear intermediate position. determines the size of cut-outs and space needed for retracting mechanism and (4) That the landing gear itself be capable of absorbing the prescribed landing retracted landing gear. In order to house the retracted landing gear and retracting loads without damage to the fuselage or airplane. mechanism, a thick wing section is necessary. For this reason, on a tapered wing (5) That maintenance of the landing gear be readily accomplished. monoplane the landing gear is retracted in close to the fuselage to make use of (6) That the landing gear be lowered or dropped with reasonable speed. the maximum wing thickness available. This eliminates the use of outboard wing (7) That the wing or fuselage openings be closed by means of doors, fairing, fuel tanks. On this type of airplane the landing gear cannot be retracted into the etc., on retraction of the landing gear. center section of the wing, due to the large cut-out necessary in the wing at the Retractable landing gears may be classed as follows: wing attachment fittings, which are at a critical section of the wing. With this type (1) Fuselage type, with landing gear retracting into the fuselage and the mecha- of retractable landing gear it is most important to close tip the wing openings on nism housed within the fuselage. retraction by means of doors, plates, or fairing. An excess weight is added due to (2) Wing type, with landing gear folding up into the wing and the mechanism addition of the doors, plates, or fairing for closing the wing openings, due to an located in the wing. inefficient wing structure necessary to house the retracted landing gear and due to (3) Engine nacelle type, with landing gear retracting back into the wing, engine the addition of the retracting mechanism. Due to the wheels being forward of the nacelles, or both, of multiengine monoplanes with the engines located outboard of front spar, the axis of rotation or hinge line must be at an angle to the chord line in the fuselage. order to fold the landing gear up between the spars. Where a slide tube or screw is The fuselage type of landing gear may be either a single-wheel or a 2-wheel used, unless the spars are far enough apart, ball and socket joints must be used on landing gear. The single-wheel landing gear does not require as wide a fuselage as the the struts to prevent interference with the rear spar, due to the landing gear swinging 2-wheel on retraction; but requires for stabilization the use of two small auxiliary at an angle to the chord, since, with pin joints, the retracting mechanism must be in wheels, which must also be retracted into the fuselage. This requires the use of two a plane perpendicular to the axis of rotation or hinge line. This type of retractable retracting mechanisms-one for the main wheel and one for the two auxiliary wheels. landing gear permits the use of monocoque fuselage. This means excess weight and more fuselage cut-outs. The engine nacelle type of retractable landing gear is restricted to monoplanes The fuselage type is best suited for a high-wing monoplane. The distance from with multi-engines located outboard of the fuselage. The engine nacelles are used the wing to ground on a high-wing monoplane prohibits the use of the wing type, advantageously to wholly or partially house or fair-in the retracted landing gear. since wing cut-outs and length of retracting mechanists depend on the landing gear This landing gear on a low-wing monoplane swings about the fitting on the front length. This would give a very poor and inefficient installation. The fuselage-type spar back into the wing in front of the rear spar. The wheels of this type landing landing gear occupies useful and valuable space at an important section of the fuse- gear cannot be fully retracted into the wing due to the size or diameter of the wheels lage near the center of gravity of the airplane. In cargo or bomber airplanes cargo and the depth of the wing section in front of the rear spar, for example, the YB-9, or bomb space is valuable near the center of gravity. In other types of airplanes this XB-907, and YO-27. In a high-wing monoplane with the engines located below place is important for the fuel tank and military equipment location. A large fuse- the wing, as on the XO-35, the landing gear swings back behind the engine. The lage section is necessary in order to house the landing gear and mechanism. The size protruding wheels on the high and low wing monoplane installation are used to of the fuselage section depends on size of wheels, landing gear length, and length advantage for protection of the airplane against possible failure of retracting mech of retracting mechanism. The length of retracting mechanism is dependent on the anism and possible failure of pilot to drop the landing gear. The design is such that 296 Chapter 3: The Design Revolution Document 3-15(a-e) 297

sector type does not require much space; but means must be provided for locking in the exposed part of the wheel strikes the ground first, thus protecting the airplane position and stops must also be provided. against serious damage. For this reason the retracted wheels are not fully faired, thus The second group makes use of cables and pulleys for retraction and lowering saving the added weight of streamlined fairing. This landing gear is best suited for of landing gear. Cables are easily adapted to an airplane, but the system of pulleys heavy cargo, bombing, or 3-place observation monoplanes and does not take up any and cables must be complicated due to the necessity of cables for releasing lock on valuable fuselage or wing space. Only small wing cutouts are necessary. Due to its landing gear for retraction as well as the double system for lowering and retracting short length, this landing gear is not heavy, and its weight compares very favorably landing gear. Tension must be maintained on the free cable on lowering to prevent with the non-retracting landing gear. It permits use of monocoque fuselage. For over-running of cable on drum and snagging of cable on parts of the airplane. Stops comparison of the retractable landing gear with the non-retractable landing gear the and catches must be provided to lock the landing gear in down position and to take following table is given: the required loads on landing. Care must be used in this installation to prevent Retracting mechanisms may be classed as follows: chafing of the cables. The double system of cables insures more positive locking (1) Positive acting, as the screw or gear type. of landing gear in lowering against any air resistance, as it does not rely on gravity. (2) Cable operated for retraction and lowering. Frequent inspection of the cables is necessary. (3) Cable operated on retraction only, but dropping by gravity. The third class depends on cables for retraction and on gravity for lowering. It (4) Hydraulic or oleo action. is essential to have tension in the cable at all times on dropping of landing gear to The first group is positive in action and is very suitable for heavy airplanes. The prevent the slack cable from snagging or overrunning the drum. This type gives a screw has the advantage of self-locking the landing gear in any position. It can be lighter installation than the second type. The cable system must also include means readily adapted for any speed, but means must be provided for lowering the land- of releasing the lock or catch on retraction. Air loads on some landing gears may be ing gear fast and retracing the landing gear with a reasonable speed. The screw type such as to resist the dropping of the landing gear and prevent the locking of it in requires a long screw, which in turn requires much fuselage or wing space. The gear- place. For this reason the single cable system is not recommended. 298 Chapter 3: The Design Revolution Document 3-15(a-e) 299

The fourth class, that is, hydraulic or oleo type, gives a good installation due 6E cantilever landing gear is 6.9 percent of the gross weight, or an increase in weight to the absence of moving parts. The tubes from the cockpit to the cylinder can be of 1.1 percent due to the installation of the cantilever landing gear with streamlined easily located to suit the airplane. A long cylinder is required to obtain the desired wheel fairing. The 0-25C and the YO-31 also form a good comparison, as their travel of landing gear. For this reason the use of this type is limited to space avail- gross weights are practically equal and their design landing load factors are the same. able for the installation. With the tubes properly fastened and supported and with The 0-25C split-axle landing gear weighs 4.8 percent of the gross weight, and the frequent inspection of tubes and connections, there is little danger of tube breakage YO-31 cantilever landing gear weighs 5.8 percent of the gross weight, which is 1.0 which would render the retracting system useless. This type is easily adapted for any percent heavier due to the use of wheel fairing and cantilever landing gear. weight of landing gear and is positive in action in either lowering or retraction of The YO-1G uses a split-axle type of landing gear and streamlined wheel fairing. landing gear. An excess weight is added due to the addition of pump, valves, hydrau- The weight of this type of landing gear is 5.6 percent of the gross weight compared lic unit, and tubing. to 5.2 percent for the same airplane without the wheel fairing, or an increase of 0.4 Table 1 shows the actual weights of various types of landing gears as used on percent due to the addition of wheel fairing. present Army airplanes. The retractable landing gear weights are grouped in their The weight of the XA-8 fixed streamline landing gear, which is attached to the proper class. The retractable landing gear weights of the fuselage or wing type do wing, is very heavy, that is, 8 percent of the gross weight. This is due to the weight not offer a fair comparison with the other types of landing gears, as the retractable of the fairing (50 pounds) and the provisions for synchronized guns, etc. landing gear weights of these two classes do not include the excess weight neces- Table 2 shows the high speed of various airplanes and models with retractable sary in the inefficient structure of the wing or fuselage due to cut-outs, nor the landing gears, with landing gears removed, with cantilever landing gears, with low- excess weight due to a larger fuselage or thicker wing section necessary to house wing monoplane landing gears, and with split-axle type of landing gears. The figures the retracted landing gear and retracting mechanism. The retractable landing gear show a very high increase in high speed due to the retraction of the landing gears. In weights of the engine nacelle type with landing gears attaching to engine mount the cases of the XP-900, XO-40, XB-907, and YlC-23, the retraction of the landing do not include the heavier engine mount or the weight of added cowling which is gear increases the high speed about 30 percent. These retractable landing gears have charged to the engine. In using the figures of Table 1 the design landing load factors an exceedingly high drag when lowered, due to the large fuselage or wing openings, must be taken into consideration. due to no streamline fairing of the struts or wheels, and due to exposed fairing or Of the six Army airplanes using retractable landing gears, four employ the use of doors for closing the openings being open. cables for retraction and lowering of the landing gear. The XO-35 uses the hydraulic Flight tests of XB-907 were made for high speed with and without landing system for retraction, and at first glance seems rather light in weight; but, due to the gear shielding with landing gear down and with landing gear retracted. The high engines being located below the wing, the landing gear is very short. However, the speed with wheels retracted was the same (195.8 miles per hour) without landing landing gear is very simple, and the only added weight is the oleo unit, tubes, valve, gear shielding as with landing gear shielding, but with wheels down the high speed and pump. It is very accessible and provides for easy maintenance. increased 23 miles per hour by removing the shielding. On this airplane the shield- The YB-9 uses the screw type mechanism for retraction . It is heavier than the ing is a detrimental clue to the excess weight and increased drag with wheels down XO-35 landing gear, due partly to the wing location with respect to the ground, requiring a longer distance for take-off. requiring a longer landing gear. However, it is lighter than the landing gear of the These speed increases are not true indications of the actual gain in speed of a XB-907, using cables for retraction. retractable landing gear over a well streamlined non-retractable landing gear. Wind The cantilever landing gears from Table 1 seem to be heavier than the retract- tunnel tests on the XA-8 and XP-16 models show an increase of only 9.8 and 7.7 able landing gears at first glance, but the retractable landing gear weights do not percent, respectively, for the removal of these streamlined landing gears. This is the include excess fuselage or wing weight. The cantilever landing gear weights are true, ideal condition of retraction, but it is never realized in practice since perfect closing so they may be easily compared with split-axle type landing gear weights. Both up of openings is not possible and since changes must be made in the fuselage or monoplanes and biplanes using cantilever or non-retractable landing gears are listed wing to accommodate the retractable landing gear and its mechanism. It is safe to in Table 1, but they may be compared directly, as these landing gears attach directly assume that the actual speed increase of a retractable landing gear is not greater than to the fuselage and are independent of wing location or the type of wing. The P-6 6 percent over the split-axle type or low-wing monoplane landing gears. and XP-6E are good examples for comparison as they are almost identical, with A good comparison of high speed can be made between the cantilever and the exception of the landing gear, and have same load factor. The weight of the P-6 split-axle type of landing gears by taking the P-6A and P-6E airplanes which have split-axle landing gear is 5.8 percent of the gross weight, and the weight of the XP- same engine. The P-6E with a cantilever landing gear weighing 6.9 percent of the 300 Chapter 3: The Design Revolution Document 3-15(a-e) 301 gross weight has a high speed of 194.5 miles per hour at 5,000 feet, and the P-6A cal advantage throughout its entire travel. The use of cables and slide tube or screw with split-axle type of landing gear weighing 5.8 percent of the gross weight has a mechanism is recommended for this landing gear. high speed of 176.8 miles per hour at 5,000 feet, or a gain of 10 percent in high The landing gear of Figure 5 is also of the first class. The struts form a paral- speed due to the use of cantilever landing gear which weighs 1.1 percent of the gross lelogram and the wheels are kept in a vertical plane in any position. A hydraulic weight more than the split-axle type. mechanism is used for retraction. This landing gear requires a wide fuselage and From actual flight tests on the YO-31 with cantilever landing gear a high speed large fuselage cut-outs. However, it does not require a deep fuselage. increase of 2.2 percent was obtained by use of wheel fairing with an increase of 0.8 Figures 6 to 9, inclusive, are landing gears of the second-class retraction into the percent in gross weight. From flight tests on the O-1G (split-axle type landing gear) wing. Figure 6 shows one of these landing gears using a worm gear and gear sector a high speed increase of 2.0 percent was noted due to the use of streamlined wheel for retraction. A stop and catch or locking device must be used to prevent landing fairing, which increased the gross weight by 0.5 percent. The high speed increase loads being taken out by the mechanism. Means for releasing the locking device of the XP-20 (split-axle type of landing gear) obtained from the flight tests was 1.0 must also be provided for retraction. This mechanism does not require a thick wing percent by the use of wheel fairing. This data indicates that streamlined wheel fair- section. Figure 7 is similar to Figure 6, except cables are used entirely for retraction ing is essential for high performance of airplanes with non-retractable landing gears, and lowering of the landing gear. Figure 8 shows a landing gear which uses a slide as a large increase in high speed results from a small increase in landing gear weight tube and cables for retraction. A latch, or locking device, and a stop is also necessary due to wheel fairing. to take the landing loads. This mechanism does not require a thick wing section. It Table 2 shows that the YlC-17 with wire-braced landing gear and wheel fairing also has a good mechanical advantage in any position of the landing gear. The land- is 6.3 percent faster than the YlC-12 with split-axle type of landing gear with an ing gear shown in Figure 9 has good possibilities. The pull of the cable on retraction increase in landing gear weight of 1.1 percent of the gross weight. Since the high is changed from the slide to the landing gear proper to the best advantage as the speed on the YlC-17 is not official and since the engine of the YIC-17 is 50 horse- slide pulley passes the fixed pulley. Thus the first pull of the retracting cable breaks power greater than that of the YlC-12, no information can be gotten from these the slide loose and as the slide passes the fixed pulley the cable leaves the slide pulley figures. and the landing gear is lifted directly, thereby giving a straight and direct pull on the Figures 1 to 3, inclusive, show several possible methods of retracting a single landing gear. This landing gear requires a thick wing section. All of these landing wheel landing gear. These landing gears use two small auxiliary wheels for stabiliza- gears must swing about an axis or hinge line which is at an angle to the chord line tion on the ground. The landing gear of Figure 1 does not fully retract, only by the in order for the landing gear to retract between the spars. For this reason, where the amount of oleo travel. It employs the use of a good streamlined fairing for the wheel. landing gears are not far enough apart to permit the landing gear to swing in the The stabilizing arms are retracted vertically into the fuselage and are so shaped to same plane as the mechanism, ball and socket joints must be used. close up the fuselage openings on retraction. Figures 2 and 3 show two different The landing gear shown in Figure 10 comes in neither the first or second class methods of retracting the auxiliary wheels in addition to showing the retraction landing gear, but is a cross between the two. The landing gear folds up into the wing of the large wheel. In Figure 2 the large wheel retracts up and back in the fuselage and the retracting mechanism is housed in the fuselage. It has a good mechanical to permit more room in the fuselage for the retracting mechanism. The auxiliary advantage throughout its travel. It has the disadvantage of cut-outs at a critical sec- wheels swing back into the fuselage and fairing is so shaped to close up the fuselage tion of the wing; that is, the point of attachment of the wing to the fuselage, and openings. Figure 3 shows the large wheel retracts vertically into the fuselage as well also eliminates the use of fuel tanks in the center section of the wing. This landing as the stabilizing arms, which also close up the openings in the fuselage. The land- gear does not require a wide fuselage section nor does it require a thick wing section, ing gears of Figures 2 and 3 use a hinged cover to close up the opening in fuselage as it retracts into the thickest section of the wing. when the main wheel is retracted. Any one of the retracting mechanisms may be used, depending on the fuselage space available. These types of landing gears do not CONCLUSIONS require as wide a fuselage as a 2-wheel landing gear. The use of a 3-blade propeller As the result of this study, it was found that a good installation of a retractable can be used advantageously to lower the fuselage, thereby shortening the landing landing gear to an existing airplane cannot be made efficiently or economically gear length. without a complete redesign of fuselage or wing. The landing gear shown in Figure 4 is very suitable for a high-wing monoplane, The wing and fuselage types of retractable landing gears in present designs are and does not require a complicated mechanism, but has the disadvantages of Class complicated, heavy, and unsatisfactory in operation. For these reasons the retract- I landing gears as stated on page 2. It is positive in action and has good mechani- able landing gears shown in Figures 1 to 10, inclusive, are proposed and show good 302 Chapter 3: The Design Revolution Document 3-15(a-e) 303

possibilities for future designs. The fuselage type of retractable landing gear does not rence Sperry built an amphibian flying boat which is believed to have had the first readily permit the use of monocoque fuselage structure. practical retractable gear. As a result of the study of the nacelle type of retractable landing gears, it is However, these were special purpose planes. On normal civilian and military believed that this landing gear will give far better high-speed performance than a planes of that time the drag of the landing gear compared with the total drag of the cantilever or split-axle type of landing gears for multiengine monoplanes such as airplane was not sufficient for a retractable gear to give enough increase in speed to bombers, cargo, or 3-place observation airplanes. This type of retractable landing warrant the additional complication. For the next ten years the problem received gear does not interfere with use of monocoque fuselage structure. The use of a can- little attention. Today most high speed planes are “cleaner,” so that the landing gear tilever landing gear on these airplanes would require a heavy landing gear and would drag is a greater proportion of the total. Thus, the gain by the elimination of land- increase the drag of the airplane. The use of a screw or hydraulic retracting mecha- ing gear drag is likely to be greater, though the value of tail wheel retraction is still nism gives a reliable and good installation for the engine nacelle type of retractable doubtful. landing gear. As a rule, where the genera arrangement allows a reasonably efficient retractable It was also found that fairing or shielding on the nacelle type of retractable landing gear it has been found that, compared with a similar design with fixed wheel landing gear may not increase the high speed with wheels retracted and that it may type undercarriage, the speed is increased only 3-4 per cent at 130 m.p.h. Because decrease the high speed materially with wheels down. Thus the excess weight of faster planes are generally cleaner the increase grows to 6-7 per cent at 150 m.p.h. fairing or shielding is of no advantage, but a detriment due to increased drag with and at 165-170 m.p.h. to about 10 per cent. One well known low wing design had wheels down, requiring a longer distance for take-off. a high speed of about 175 m.p.h. with a streamlined external gear, while with the The results of this investigation also show that the cantilever landing gear with landing gear fully retracted the speed was increased some 25 m.p.h., or 14 per cent. streamlined wheel fairing gives the best installation for monoplanes or biplanes in It is possible that as much as 20 per cent increase in speed might result at 185-195 the single-engine class, such as pursuit, 2-place observation, attack, or light cargo m.p.h. from retracting the landing gear on a very clean design. On designs where airplanes. This installation permits the use of a small fuselage, which is of vital the wing bracing is combined with a streamlined landing gear, as on may low wing importance in a pursuit, attack, or observation airplane for vision and high-speed racers or the Bellanca Airbus, the gain due to retraction would of course be debatable. performance; and it requires no fuselage or wing space, which is valuable in these At the speeds achieved today the exact relative merits of the streamlined external airplanes for fuel tank or military equipment locations. This landing gear can be gear (whether or not it is part of the wing bracing) and the retractable type depend easily adapted to existing airplanes without any major changes or alterations, includ- much upon the ingenuity of the designer and the purpose, type and size of the ing monocoque fuselage structures. airplane. It is believed that the cantilever landing gear with wheel fairing will give a better high-speed performance than the low wing monoplane landing gear and the wire Space braced landing gear with a lighter installation. On some type insufficient space is available in the wing or fuselage to retract This study proves conclusively that the use of streamlined wheel fairing is the landing gear. Fundamentally, for high speed flight it is believed better to increase important for high-speed performance for non-retractable landing gears with only the frontal area slightly to provide the needed space. Thus all parts are concentrated a small increase in gross weight. into one mass that can be given the best possible shape, rather than have a number of separate units with the accompanying interference between them and the possible disturbances where they meet. This problem is closely tied in with problems of Document 3-15(d), Richard M. Mock, “Retractable Landing Gears,” where to house the wheels and will be discussed later. Aviation 32 (February 1933): 33-37. Weight Though retractable landing gears for landplanes have only come into popular use in the last few years, the idea is by no means new. In 1876 Penaud and Gauchot An increase in speed with the same power means less flying time, hence less fuel patented a design of an airplane which had a front landing chasses of the wheel, and oil required for a given flight distance. The saving in weight of fuel and oil goes all retracting into the fuselage to reduce the air resistance. However, as far as can to offset the additional weight of the retracting mechanism, linkages and accompa- be learned the idea never took a practical form until after the War, when Dayton- nying increase in structural weight. Wright built a high wing cantilever monoplane for the 1920 Gordon Bennett race An example might help to illustrate this point. Assuming an airplane having with the wheels retracting flush in the sides of the fuselage. The year before Law- 8,000 lb. gross weight, 3,000 lb. useful load, and cruising at 150 m.p.h., the addi- 304 Chapter 3: The Design Revolution Document 3-15(a-e) 305

tion of a retractable mechanism might increase the weight empty 60 lb., and reduce An increase of 5 per cent in cruising speed, however, would mean that a flight of the useful load by the same amount. The fuel consumption might be assumed to be a given distance could be made in 4.8 per cent less flying time and hence at a saving 35 gal. and oil consumption 3 gal. per hour, or a total of 230 lb. per hour. A 10 per of 4.8 per cent minus the added costs just tabulated. A 15 per cent increase would cent increase in speed would reduce the time for a given flight 9.1 per cent, saving cut the time 13 per cent. It is thus apparent that in a typical case the provision of a about 21 lb. of fuel and oil per hour (neglecting that slightly smaller tanks could retractable gear ought to reduce the operating expenses by from 3 to 12 per cent of be used.). Therefore for a 450 mile flight, if this increase in speed is achieved, the the direct flying costs, or from 1.0 to 3.6 per cent of the total cost, depending upon additional weight of the retracting gear would have no net effect on the pay load. the gain in speed. For a transport carrying from 8 to 12 passenger and crusing at This result is typical, whatever the size of the plane. 160 m.p.h., a retractable gear ought to cut total costs about $1-$2.50 per hour or In general a fixed landing gear weighs 6-9 per cent of the empty weight of the 0.625—1.56 cents per mile. airplane or 10-14 per cent of the useful load, though this may vary as much as 7-17 per cent. A retractable landing gear increases the landing gear weight 10-20 per cent, Danger of failure reduces the useful load 1-3 per cent. Airplanes of about 10 lb. per hp. Loading use Another factor to be considered is the danger of the landing gear not being in 7-11 per cent of their useful load per hour in fuel and oil. Thus, with a three hour the proper position at the time of landing. This may happen from any one of four range the additional weight would be offset by the fuel and oil saving if the speed reasons: (1) A structural failure of a landing gear part or the actuating mechanism, were increased 5-7 per cent. For a two hour flight, a 10-13 per cent increase would due to faulty design, materials or workmanship; (2) fouling of the wheel bracing be sufficient. linkage, or actuating mechanism by ice or mud; (3) pilot’s neglect to lower the wheels, which an be eliminated by proper warning devices; or (4) a mechanism that Cost is too slow to allow the wheels to be lowered in a forced landing Outside of the greater value of a faster service, an increase in speed often means As far as can be determined there has been no instance of a personal injury due that fewer planes are needed to keep a given frequency of schedule. In addition, the to a retractable landing gear functioning improperly. In a forced landing in a small flying time to cover a given distance is less, meaning a reduction of all costs pro- or bad field, especially with passengers a “wheel up” landing is definitely an advan- portional to flying time, such as fuel and oil, flying crews pay and maintenance and tage as the speed is slightly lower, the run considerably shorter and the possibility overhaul of airplane and engine. of nosing over practically eliminated. One of the leading operating companies has Consider a conservative approximation of the additional expense due to a found it desirable to make forced landings with wheels retracted, as the expense of retractable gear. Assuming that the initial cost of the retractable landing gear is subsequent repairs is usually less than if the wheels were completely or partly down. $500-$1,500, depending on the size, type and purpose of the plane, there will be In the past two years a great number of wheel up landings have been made with an increase in cost of depreciation of 10 cents to 30 cents per hour based on 5,000 various privately and commercially operated planes, and invariably they resulted in hours for the life of the airplane. Though some operators claim a retractable gear less than $1,000 damage and more often $300 or less. The greatest damage usually is cheaper to maintain than a fixed gear, with streamlining, assume an additional experienced is when the wheels are in a partly retracted position. The extent of the maintenance cost of 3 cents to 10 cents per flying hour making a total additional damage due to a wheel up landing is usually limited to the propeller and the bottom expense of 13 cents to 40 cents per flying hour. covering of the fuselage. Though in most instances the insurance costs would not be increased, it seems To insure proper functioning of a retractable landing gear the mechanism should reasonable to assume a possible increase of 1 to 2 per cent in both crash and pas- be designed to be simple, easy to service and maintain, and well protected against senger liability insurance. From the Post office Department figures, insurance rep- being fouled by foreign material thrown up by the wheel during take off. With the resents 6.5 per cent of the total operating expense. Crash and passenger liability are wheel opening in the bottom of the wing or fuselage the mechanism should not be assumed to constitute about 40 per cent of the total insurance or 2.6 per cent of the exposed to mud, which might freeze solid after the gear has been retracted. In plan- total flying expense. If this were increased 1 to 2 per cent due to the retractable gear ning the disposition of the landing gear members, care should be taken to prevent the total increase in operation expenses would be only 0.026 to 0.052 per cent. struts from being forced into the cabin if a landing is made with partially retracted Using the most probably figures within the above range in each case, and adding wheels. The actuating mechanism, especially cables, should not take any landing an imaginary plane of strictly up to the minute design and correspondingly lowered loads and the linkage must be so laid out that it never approaches dead center closely operating costs, the following table shows the increase in cost due to the addition enough to require great operating force. The mechanism should not be reversible as of a retractable gear. (Only a portion of the “direct operating expenses” are directly the “down position” is approached (unless sole dependence is to be placed on a lock proportional to flying time so only 30 per is used.) to hold the wheels in landing position). 306 Chapter 3: The Design Revolution Document 3-15(a-e) 307

Wheel location The best method and location for housing the landing gear varies with the type of plane. The gear can be retracted into one of five places :into the wing, the fuselage, and engine nacelle of a muti-engine design, an external fairing, or an exposed position where the frontal area or interference drag is reduced. In deciding which one to use, the aerodynamic effect of the wheel opening and the partly retracted gear during take off and landing immediately invites consideration. A disturbance of the flow by the opening might affect the stability, while a wing opening might affect the airfoil characteristics. These aerodynamic effects seem to be negligible however in most designs. Recent full scale wind tunnel tests by the N.A.C.A. on a Lockheed monoplane show that a wheel opening in the lower surface of the wing does not affect the flying characteristics when the wheels are extended. (Reference 2) Until a year or two ago, practically all American air planes were biplanes or high wing monoplanes. As these types have inherently less drag than the low wing monoplane their general arrangement should be more desirable. (References 3, 4 and 5) The ease of retracting the gear in the low wing monoplane is apparently one of the main reasons for its increasing popularity. It is interesting that with the same power and loading a low wing monoplane, which gained 14 per cent in speed by fully retracting the landing gear, was only 3-4 per cent faster than a high wing version of the same plane with a fixed streamline, single strut, tie rod braced gear indicating that the high wing model would have been considerably faster with a retractable gear. However, as it is exceptionally difficult to retract the landing gear into the wing of a high wing type the fuselage is a more promising location for single engine designs. There have been many successful designs with the wheels retracting either flush into the sides of the fuselage or into the bottom. An interesting mechanism adaptable for retracting the wheels into the sides of the fuselage is shown in the illustration taken from a patent held by L. R. Grumman. A deviation from a true parallelogram is made to provide “toe in” for landing and taxing. Another type of parallelogram mechanism was used on the Great Lakes amphibian of 1929. The member parallel to the wheel was an oleo carrying bending from the axle. A diagonal strut hinged at the top of the oleo went downward and inward to a point in the hull. By raising the inner end of this member the wheel was raised and swung inward into a recess in the side of the hull. To have sufficient tread and ground clearance on a high wing monoplane with the wheels retracting into the bottom of the fuselage it is usually necessary to either build out small stubs as on the Stinson R-3 or to have a wide fuselage as on the Burnelli. The mechanism of the Stinson is illustrated herewith. Burnelli uses a sort of curved track to give a specific path to the upper end of one member of a tripod strut arrangement. There are an infinite number of other variations with shortening or retracting struts, folding struts, etc. To retract the wheel into a thin wing as on a biplane or an externally braced monoplane would usually necessitate and increase in wing thickness to house the